Early Invasive vs Conservative Treatment Strategies in Women and Men With Unstable Angina and Non–ST-Segment Elevation Myocardial Infarction: Title and subTitle BreakA Meta-analysis FREE

Although an invasive strategy is frequently used in patients with unstable angina and non–ST-segment elevation myocardial infarction (NSTEMI), data from some trials suggest that this strategy may not benefit women. Despite an overall benefit for patients randomized to an invasive strategy in the FRISC II (Fragmin and Fast Revascularisation During Instability in Coronary Artery Disease) and the RITA 3 (Randomized Intervention Trial of Unstable Angina) trials,^{1 - 3} subgroup analyses from these studies showed that an invasive strategy may be associated with a higher risk of death or MI in women.^{4 - 5} In contrast, an early invasive strategy was associated with improved outcomes in both men and women in the TACTICS-TIMI 18 (Treat angina with Aggrastat and determine Cost of Therapy with an Invasive or Conservative Strategy–Thrombolysis in Myocardial Infarction 18) trial,⁶ with particular benefit observed in high-risk women identified by an elevated serum troponin. Thus, the benefit of an invasive strategy in women remains unclear. However, individual trials have not been large enough to explore outcomes reliably within subgroups. For this reason, we conducted a collaborative meta-analysis of randomized trials to examine the benefits and risks of an invasive strategy in women vs in men with non–ST-segment elevation acute coronary syndromes (NSTE ACS).

A computerized literature search was conducted from 1970 through April 2008 of the MEDLINE and Cochrane databases by using search terms that included invasive strategy, conservative strategy, selective invasive strategy, acute coronary syndromes, non-ST-elevation myocardial infarction, and unstable angina. No other search restrictions were applied. In addition, we contacted experts in the field and hand-reviewed abstracts from major cardiology meetings.

Trials considered for inclusion were those that enrolled individuals with a diagnosis of NSTE ACS (unstable angina or NSTEMI) and randomized patients to either an invasive or conservative treatment strategy. An invasive strategy was defined as the referral of all patients with NSTE ACS for coronary angiography followed by revascularization if deemed appropriate. A conservative treatment strategy was defined as a primary strategy of pharmacological management and subsequent coronary angiography reserved only for those patients with recurrent symptoms of unprovoked ischemia or objective evidence of inducible ischemia on noninvasive testing.

Trials were excluded from participation if they primarily enrolled patients with stable angina or acute ST-segment elevation myocardial infarction (STEMI). Trials were excluded if data were not previously peer-reviewed, fibrinolytic therapy was to be administered in all patients, inadequate randomization was performed, coronary angiography was required before patient enrollment, or reliable sex-specific outcome data were unavailable.

The principal investigators (W.E.B., E.B., C.P.C., R.J.W., K.A.A.F., P.A.M., R.S., L.W.) for each participating trial provided the sex-specific incidence of death, nonfatal MI, and rehospitalization with ACS, and the composites of these end points. Outcomes were collected from (1) randomization to hospital discharge, (2) hospital discharge to the end of 12 months of follow-up, and (3) randomization to the end of 12 months of follow-up. For the TACTICS-TIMI 18 trial,⁷ follow-up data were available through 6 months of follow-up. Outcomes were also substratified by the presence or absence of cardiac biomarker elevation (creatine kinase MB or troponin) and ST-segment deviation on electrocardiogram. Patients without available biomarker or ST-segment deviation data were excluded from the respective high-risk subgroup analyses. Outcomes were collected for patients in each treatment group who ultimately underwent percutaneous coronary intervention or coronary artery bypass graft (CABG) surgery.

For the purpose of the current meta-analysis, nonfatal MI (both procedure-related and non–procedure-related) and rehospitalization with ACS were defined as reported in each of the individual trials. In addition, as per their original manuscript, the investigators from the ICTUS (Invasive vs Conservative Treatment in Unstable Coronary Syndromes) trial⁸ provided incidence rates for nonfatal MI by applying the definitions of MI as used in the TACTICS-TIMI 18 and FRISC II trials.^{1 ,7}

For patients randomized to an invasive strategy, data were collected on the extent of coronary artery disease (CAD) at the time of coronary angiography. A diseased vessel was defined as a major epicardial vessel or CABG with at least 50% stenosis. In the ICTUS trial,⁸ data were provided for at least 70% stenosis in a major epicardial vessel or CABG.

All of the collected data were verified by each of the participating investigators (W.E.B., E.B., C.P.C., R.J.W., K.A.A.F., P.A.M., R.S., L.W.). When required, participating investigators performed new analyses to generate tabular data from the original trials.

A meta-analysis was performed of the relative odds for each trial, both overall and stratified by sex, and based on random-effects models using the method by Der Simonian and Laird.⁹ Results are presented as odds ratios (ORs), with their 95% confidence intervals (CIs). Event rates for each trial individually and the pooled data are presented as frequencies. Heterogeneity between sex and between high-risk subgroups was assessed by calculating the OR for the interaction term between treatment strategy and subgroup for each individual trial, combining these ORs in random-effects models, and calculating P values for the combined interaction effect. For the biomarker analyses, the meta-analysis was restricted to those 5 trials that enrolled patients both with and without biomarker elevation (TIMI IIIB [Thrombolysis in Myocardial Infarction],¹⁰ MATE [Medicine vs Angiography in Thrombolytic Exclusion],¹¹ FRISC II,¹ TACTICS-TIMI 18,⁷ RITA 3² ) to enable us to examine the relative benefit of each strategy across these subgroups.

With the available sample size and event rates, our meta-analysis had 80% power to detect a 33% difference in the relative odds of death, MI, or rehospitalization with ACS in the invasive vs conservative groups between men and women. Corrections were not made for multiple hypothesis testing given the exploratory nature of the analyses. All statistical analyses were performed by using Stata/SE version 9.0 (StataCorp, College Station, Texas). All tests were 2-sided, with P < .05 considered significant.

A total of 18 randomized clinical trials were identified through literature review (Figure 1).^{1 - 2 ,7 - 8 ,10 - 23} Of these trials, 10 trials were subsequently excluded from participation because fibrinolytic therapy was to be administered in all patients in 2 trials,^{15 ,17} coronary angiography was required before enrollment in 4 trials,^{12 - 14 ,16} patients were enrolled with stable angina in 1 trial,²¹ patients were randomized based on the day of the week in 1 trial,¹⁹ patients were referred in both treatment groups to undergo coronary angiography within 5 days of randomization in 1 trial,²² and 1 trial had not yet been peer-reviewed and published.²³ A total of 8 remaining trials were determined to be eligible and were invited to collaborate (Table 1).^{1 - 2 ,7 - 8 ,10 - 11 ,18 ,20} Of the 8 invited trials, all the principal investigators agreed to participate in a collaborative meta-analysis and contributed sex-specific outcome data in tabular format.

Glycoprotein IIb/IIIa inhibitors were available for use in 4 of the 8 trials^{1 - 2 ,7 - 8} and were required per-protocol in the TACTICS-TIMI 18 trial.⁷ In the ICTUS trial,⁸ all patients who underwent an interventional procedure during the index hospitalization were to be administered a glycoprotein IIb/IIIa inhibitor. All patients in the VANQWISH (Veterans Affairs Non-Q-Wave Infarction Strategies in Hospital),¹⁸ VINO (Value of First Day Angiography/Angioplasty In Evolving Non-ST Segment Elevation Myocardial Infarction),²⁰ and ICTUS⁸ trials required cardiac biomarker elevation to be eligible for trial enrollment. In the FRISC II trial,¹ data on rehospitalization with ACS were available in patients enrolled in Denmark and Sweden, representing 97% of the study population.

These trials enrolled a total of 10 412 patients; sex-specific baseline characteristics are displayed in Table 2. Across all trials, the weighted mean age of patients was 64.1 years in women and 61.3 years in men. Women were significantly more likely to have baseline comorbidities, including diabetes mellitus, hypertension, and hyperlipidemia. Men were significantly more likely to be active smokers and have a prior history of MI. The incidence of ST-segment depression was similar in both men and women; however, women were significantly more likely to have evidence of T-wave inversions on electrocardiogram. Men were significantly more likely than women to have elevated cardiac biomarkers (creatine kinase MB or troponin) at the time of randomization.

The extent of CAD was compared in men and women randomized to an invasive strategy and who underwent coronary angiography. Women were more frequently found to have no significant CAD (<50% stenosis) at angiography than were men (24% vs 8%, P < .001) (Table 2), even if restricted to only those patients with ST-segment deviation on electrocardiogram (16% vs 6%, P < .001) or to those patients with elevated biomarkers (14% vs 6%, P < .001). Conversely, men were more frequently found to have 3-vessel or left main disease compared with women (35% vs 23%, P < .001).

Data on the incidence of death, MI, or rehospitalization with ACS were available in 10 150 patients, including 3075 women and 7075 men. In the overall population and irrespective of sex, the composite end point of death, nonfatal MI, or rehospitalization with ACS occurred in 1075 of 5083 patients (21.1%) in the invasive group and 1313 of 5067 patients (25.9%) in the conservative group by the end of 12 months of follow-up (OR, 0.78; 95% CI, 0.61-0.98) (Figure 2 and Table 3).Compared with the conservative strategy, the OR for death with an invasive strategy was 0.97 (95% CI, 0.71-1.32), the OR for recurrent nonfatal MI was 0.84 (95% CI, 0.63-1.12), the OR for death or MI was 0.92 (95% CI, 0.69-1.23), and the OR for rehospitalization with ACS was 0.68 (95% CI, 0.55-0.84) (Table 3).

The end point of death, MI, or rehospitalization with ACS occurred in 709 of 3075 women (23%) and 1679 of 7075 men (24%) when pooled across treatment groups. The OR for the composite of death, MI, or rehospitalization with ACS with an invasive vs conservative strategy was 0.81 (95% CI, 0.65-1.01) in women and 0.73 (95% CI, 0.55-0.98) in men without statistically significant heterogeneity between sex (P for interaction = .26) (Figure 2 and Table 3).

Of the individual elements of the composite end point, a routine invasive strategy was associated with a substantial 32% to 34% significant decrease in the odds of rehospitalization with ACS in both women and men (Table 3). In contrast, in women, an invasive strategy had no significant effect on all-cause mortality (OR, 1.11; 95% CI, 0.72-1.70), nonfatal MI (OR, 0.93; 95% CI, 0.59-1.45), or the composite of death or MI (OR, 1.02; 95% CI, 0.67-1.55).

In men, an invasive strategy was associated with a nonsignificant 11% lower odds of death (OR, 0.89; 95% CI, 0.58-1.35), a nonsignificant 19% lower odds of nonfatal MI (OR, 0.81; 95% CI, 0.59-1.11), and a nonsignificant 13% lower odds of death or MI (OR, 0.87; 95% CI, 0.61-1.23).

We examined clinical event rates both before and after hospital discharge to evaluate the possible existence of an early hazard in women or men randomized to an invasive strategy (Figure 3).

Before hospital discharge, an invasive strategy was associated with a nonsignificant higher incidence of death or recurrent MI in both women and men compared with those patients treated conservatively. Irrespective of sex, an invasive strategy was associated with a nonsignificant 37% increase in the odds of death or MI during index hospitalization compared with patients treated conservatively (OR, 1.37; 95% CI, 0.93-2.02). Death or recurrent MI during the index hospitalization occurred in 6.2% women in the invasive group vs 4.2% women in the conservative group (OR, 1.49; 95% CI, 0.72-3.09). In men, death or recurrent MI before first hospital discharge occurred in 6.5% men in the invasive group vs 4.8% men in the conservative group (OR, 1.33; 95% CI, 0.93-1.89, P for interaction (sex) = .45).

From the time of first hospital discharge through 12 months of follow-up, an invasive strategy was associated with a significant 28% lower odds of death or MI (OR, 0.72; 95% CI, 0.52-0.99), irrespective of sex. The OR for the composite of death or MI from an invasive vs conservative strategy was 0.71 (95% CI, 0.48-1.07) in women and 0.70 (95% CI, 0.50-0.98) in men (P for interaction (sex) = .52).

Because current guidelines recommend that high-risk features be used to help identify patients who might benefit most from an invasive strategy,^{24 - 25} we performed an exploratory analysis stratified by the presence or absence of cardiac biomarker elevation or ST-segment deviation on electrocardiogram (Figure 4). Cardiac biomarkers (creatine kinase MB or troponin) were available at the time of enrollment in 97% of men and women. Because serum troponin was only available in 69% of women and 66% of men, analyses were not restricted to this parameter. The presence or absence of ST-segment deviation was available in 99.5% of women and 97% of men.

Irrespective of sex, the OR for death, MI, or rehospitalization with ACS with an invasive vs conservative strategy was 0.59 (95% CI, 0.51-0.69) for patients with elevated biomarkers (creatine kinase MB or troponin) vs 0.79 (95% CI, 0.58-1.06) for those without (P for interaction = .18) (Figure 4). An invasive strategy was associated with a significant 32% reduction in the odds of death or MI in patients with elevated biomarkers (OR, 0.68; 95% CI, 0.56-0.82), whereas there was no reduction in death or MI in those patients who were biomarker-negative upon presentation (OR, 1.01; 95% CI, 0.79-1.28; P for interaction = .03).

In men with an elevated cardiac biomarker, an invasive strategy compared with a conservative strategy was associated with a significant 44% lower odds of death, MI, or rehospitalization with ACS (OR, 0.56; 95% CI, 0.46-0.67) and a 36% lower odds of death or MI (OR, 0.64; 95% CI, 0.51-0.81). In men without biomarker elevation, the quantitative benefit of an invasive strategy was numerically smaller for both the triple (OR, 0.72; 95% CI, 0.51-1.01) and double composite end points (OR, 0.87; 95% CI, 0.61-1.25), although the relative difference was not statistically significant (P = .09 and P = .28 for interaction, respectively).

Among women with an elevated cardiac biomarker, an invasive strategy significantly decreased the odds of death, MI, or rehospitalization with ACS by 33% (OR, 0.67; 95% CI, 0.50-0.88), which was comparable with the benefit observed overall in men. In contrast, there was less apparent benefit with an invasive strategy in women in the absence of biomarker elevation (OR, 0.94; 95% CI, 0.61-1.44), although the test for heterogeneity did not meet statistical significance (P for interaction = .36). An invasive strategy was associated with a nonsignificant 23% lower odds of death or MI in women with elevated biomarkers (OR, 0.77; 95% CI, 0.47-1.25), whereas an invasive strategy was associated with a nonsignificant 35% higher odds of death or MI in biomarker-negative women (OR, 1.35; 95% CI, 0.78-2.35; P for interaction = .08).

In addition, the presence of ST-segment deviation also identified those patients at higher risk and with modest but not statistically significant greater benefit from an invasive strategy (Figure 4).

In both the invasive and conservative treatment groups, men were more likely than women to undergo coronary revascularization, largely due to a higher rate of CABG surgery (Table 4). As an exploratory analysis, we compared outcomes between women and men by type of revascularization performed after randomization, while recognizing the limitation that treatment strategy was not randomized and that it is not possible to ascertain whether observed differences in patient outcomes can be explained by differences in baseline comorbidities or presenting features. Overall, the risk of death, MI, or rehospitalization was 23% for women and 24% for men at 12 months of follow-up. Across treatment groups, there were no significant differences in the subsequent odds of death, MI, or rehospitalization with ACS in women vs men who underwent coronary revascularization by percutaneous coronary intervention (pooled event rates, 31% vs 30%; random effects: OR, 1.14; 95% CI, 0.96-1.35). There were significantly higher odds of death, MI, or rehospitalization with ACS in women vs men who required CABG surgery by the end of follow-up (33% vs 26%; OR, 1.44; 95% CI, 1.15-1.81). Conversely, there was a nonsignificantly lower event rate in women who were treated medically without coronary revascularization compared with men (16% vs 20%; OR, 0.91; 95% CI, 0.66-1.22).

There was evidence of heterogeneity between trials for the composite end points of death or MI (heterogeneity χ² = 34.5, P < .001) and death, MI, or rehospitalization with ACS in men (heterogeneity χ² = 39.7, P < .001), and death or MI in women (heterogeneity χ² = 17.0, P = .02). Sensitivity analyses suggested that several trials contributed to this heterogeneity with the largest contribution provided by the ICTUS,⁸ VANQWISH,¹⁸ and VINO²⁰ trials. Importantly, heterogeneity did not appear to be explained by the year of publication or concomitant therapies alone. When the primary composite end point of death, MI, or rehospitalization with ACS was reanalyzed applying alternate definitions of MI for the ICTUS trial⁸ as defined in TACTICS-TIMI 18⁷ and FRISC II,¹ the heterogeneity between trials was not eliminated.

The findings of our meta-analysis suggest that both men and high-risk women, defined by elevated biomarkers of necrosis, have a comparable benefit from an invasive strategy in unstable angina and NSTEMI for reducing the odds of death, MI, or rehospitalization with ACS. In contrast, an invasive strategy does not appear to substantially benefit women in the absence of biomarker elevation and moreover may potentially increase the risk of death or MI. Our data provide evidence to support the updated American College of Cardiology/American Heart Association guidelines that now recommend that a conservative strategy be used in low-risk women with NSTE ACS.²⁴

Prior sex subgroup analyses within individual trials have yielded disparate findings.^{4 - 6} These conflicting data have led to controversy regarding the optimal treatment strategy in women with NSTE ACS. Despite an overall benefit for patients randomized to an invasive strategy in RITA 3² and FRISC II¹ trials, subgroup analyses within these trials suggested that an invasive strategy may not benefit women and may potentially increase their risk of death or MI. In the FRISC II trial,¹ the higher event rate observed in women compared with men in the invasive group appeared to be largely due to a marked increase in death (9.9% vs 1.2%) and recurrent MI (12% vs 5%) in women who required CABG surgery.⁴ To that end, there is continuing debate as to whether women are inherently at increased risk of adverse outcomes after CABG surgery, or whether this apparent increase in risk is explained by body size and comorbidities.^{26 - 31} Recent studies now suggest a more favorable prognosis than in the past for women with ACS undergoing CABG surgery.^{32 - 34}

Methods of risk stratification, including cardiac biomarkers of necrosis, ST-segment changes, and risk scores, have been previously shown to help identify high-risk patients who might benefit most from an invasive strategy.^{8 ,24 ,35} In both the RITA 3² and FRISC II¹ trials, women overall had a more favorable prognosis compared with men. In addition, women in the RITA 3 trial² were at lower risk than women in the TACTICS-TIMI 18 trial.⁵ A larger proportion of lower-risk patients may help to explain the reduced benefit from an invasive strategy in women observed in some trials and may also highlight procedural risk. To that end, in the TACTICS-TIMI 18 trial,⁵ the benefit from an invasive strategy appeared to be restricted to women with an elevated troponin T (OR, 0.56; 95% CI, 0.32-0.97), whereas there was a trend toward harm from an invasive strategy in women who were troponin negative (OR, 1.46; 95% CI, 0.78-2.72).

In our meta-analysis, we found that women with elevated biomarkers at randomization had a significant 33% lower odds of death, MI, or rehospitalization with ACS (OR, 0.67; 95% CI, 0.50-0.88) and a directionally consistent but nonsignificant 23% lower odds of death or MI with an invasive strategy (OR, 0.77; 95% CI, 0.47-1.25), benefits that were similar to the overall benefit observed in men. In contrast, women without biomarker elevation did not appear to have a significant benefit from an invasive strategy and had a nonsignificant higher odds of death or MI compared with those treated conservatively. Although the relative difference between biomarker-positive and biomarker-negative women was not statistically significant, tests for interaction are conservative and may miss a true difference when one does in fact exist. It may be challenging to detect a statistically significant difference between subgroups due to their inherently smaller sample size. Our findings emphasize the need for larger prospective trials to specifically examine the benefit of an invasive strategy in women both overall and within high-risk subgroups.

We found that regardless of sex, the benefit of an early invasive strategy was largely due to a significant reduction in rehospitalization with ACS, with more modest effects on death or MI. Consistent with prior analyses,³ we observed a signal toward an early hazard with an invasive strategy such that patients randomized to an invasive strategy had a higher rate of death or MI before first hospital discharge, but subsequently had a significant reduction in death or MI following hospital discharge compared with patients treated conservatively.

We also observed that women are significantly less likely than men to have obstructive CAD at the time of angiography despite a clinical presentation consistent with NSTE ACS. Overall, 24% of women randomized to an invasive strategy had no evidence of significant epicardial CAD at angiography vs only 8% of men. When restricted to only those patients with elevated biomarkers of necrosis, the incidence of obstructive CAD marginally increased in women and men; however, 14% of women still had no evidence of significant CAD. These findings may reflect a greater burden of microvascular disease or abnormal vasodilatory reserve in women resulting in subendocardial ischemia, coronary vasospasm, or possibly other acute disease states that mimic ACS secondary to obstructive CAD.^{30 ,36 - 37} Because only a select subset of patients in the conservative group of each trial underwent coronary angiography, it is not possible to explore the relative benefit of an invasive or conservative strategy in patients who were found to have obstructive CAD due to the risk of selection bias.

The relative paucity of obstructive CAD in women may dilute the treatment benefit of an invasive strategy and may in part explain the signal toward harm observed in some trials from an invasive strategy in women, because patients without significant epicardial disease are unlikely to obtain benefit from coronary angiography. Although biomarkers of necrosis may be useful for helping to predict the presence of obstructive CAD, consideration should be given to identify additional high-risk features or novel biomarkers that may prove to be useful for predicting the benefit of an invasive strategy in women.³⁵

There are limitations to our analysis that warrant consideration. None of the trials were individually powered to examine outcomes within specific subgroups, and subgroup analyses inherently increase the risk that the findings might be explained by chance or uncontrolled confounding.^{38 - 39} In addition, because we did not find any effect modification by sex alone, further analysis within high-risk subgroups can only be performed on an exploratory basis and these results should be confirmed with prospective trials. Because directional consistency was not observed across the individual elements of the composite end points, the secondary end points should be interpreted as hypothesis generating. Finally, as with any meta-analysis, limitations to the methods include heterogeneity between trials and the possibility of publication bias. In particular, there are differences across the trials in terms of populations, adjunctive pharmacotherapy, invasive treatment options, and outcome measures.⁴⁰ However, with regard to the latter, our findings did not qualitatively differ when alternate definitions for MI were applied to the ICTUS trial population.⁸

In our meta-analysis, we did not have access to patient-level data so it is not possible to assess whether differences in outcomes by sex can be explained by variations in baseline characteristics or presenting features. In addition, bleeding events were not collected in all trials, so we cannot ascertain whether women may be at increased risk of bleeding following angiography or revascularization. The meta-analysis spans trial publication dates of more than 10 years. Since the publication of the first trial, there have been several advances in the treatment of patients with ACS, including the routine use of glycoprotein IIb/IIIa inhibitors, thienopyridines, and evolutions with stent and surgical technologies. The use of these agents may differ both between trials and treatment groups. Moreover, the proportion of patients undergoing angiography and revascularization in the conservative groups varied widely across trials and may have contributed to the observed heterogeneity. More intensive background medical therapy may be present in more recent trials. However, importantly, heterogeneity between trials did not appear to be explained by publication date or concomitant therapies alone.

This is to our knowledge the first large-scale and collaborative meta-analysis that examines the association between sex and outcomes with an invasive or conservative treatment strategy in NSTE ACS across all of the major randomized trials. Combination of these data enabled us to explore the association of sex with outcomes both overall and within high-risk subgroups, whereas individual studies may be insufficiently powered in this regard. Future investigations should include novel methods for identifying women at high-risk of adverse outcomes after NSTE ACS and whose risk could be modifiable with an invasive approach.