3.1. Study characteristics
We found 4338 potentially articles, among which 14 trials involving 11,624 individuals were included (Fig.1). Treatment duration ranged from 1 to 24 months (8.61±5.77). Patients were randomized to spironolactone in 8 trials (N=1412), eplerenone in 4 trials (N=4081), canrenoate in 2 trials (N=365) and assigned 1408, 3990, and 368 patients to control groups, respectively. The EPHESUS trial [12] accounted for more than half of the patients. Two studies [25,13] did not use double-blind methods and one study [24] reported incomplete outcome data (Fig. 2). The kappa statistic showed a good agreement between reviewers (Online Resource 3). The Modified Jadad scores of trials varied from 5 to 7 points, indicating that this meta-analysis was a relatively high-quality report.
3.2. All-cause mortality
11 studies included 11,037 patients reported all-cause mortality. 532/5523 (9.63%) and 630/5514 (11.43%) were observed in treatment and control arms, respectively, with a general reduction of 16% (RR 0.84, 95%CI (0.76,0.94), P=.002, I2=0%, Fig. 3). In addition, reduction benefits of MRA were particularly evident in subgroups such as STEMI patients, treatment initiation within 3 days and (3,7) days (RR 0.62, 95%CI (0.42,0.90), P=.01, I2=0%; RR 0.70, 95%CI (0.49,1.00), P=.05, I2=0%; RR 0.71, 95%CI (0.59,0.86), P=.0004, I2=0%, Fig. 4), duration >6 months, patients with LVSD , eplerenone, and 25mg (RR 0.85, 95%CI (0.76,0.95), P=.005, I2=0%; RR 0.87, 95%CI (0.77,0.97), P=.01, I2=49%; RR 0.86, 95%CI (0.77,0.97), P=.01, I2=0%; RR 0.85, 95%CI (0.76,0.94), P=.003, I2=0%, Online Resource 4). No evidence of publication bias as suggested by funnel plot, the Begg’s test (P=0.64), and the Egger’s test (P=0.63) was observed (Fig. 5). None of the individual studies significantly influenced the pooled all-cause mortality estimates in the leave-one-out sensitivity.
3.3. New or worsening HF and deaths due to HF
8 RCTs involving 10,515 patients (10.74% in the MRA group vs 12.14% in the control group) showed a significant 14% reduction in new or worsening HF after MRA treatment (Fig. 3). Duration >6 months, patients with LVSD, eplerenone, and 50 mg subgroups using MRA respectively reduced new or worsening HF(RR 0.86, 95%CI (0.77, 0.96), P=.007, I2=0%; RR 0.87, 95%CI (0.77, 0.97), P=.02; RR 0.86, 95%CI (0.77, 0.96), P=.008, I2=0%; RR 0.86, 95%CI (0.76, 0.96), P=.008, I2=0%, Online Resource 4). The prevention effect of MRA for death from HF estimated by RR was 0.78 (95%CI (0.62,0.99), P=.04, I2=0%, Fig. 3). The EPHESUS trial [12] provided weights of 81.1% and 83.9% for new or worsening HF and HF mortality, respectively. RR excluding it resulted in no statistical significance: from (0.86, p=.007) to (0.86, p=.23); (0.78, p=.04) to (0.63, p=.12), respectively.
3.4. Cardiovascular deaths and hospitalizations
Pooled data showed that cardiovascular deaths reduced by 16% (RR 0.84, 95%CI (0.74,0.94), P =.003, I2=0%, Fig. 3). MRA groups (n=452/5294) had a greater reduction than control arms (n=537/5193). The analysis showed that MRA treatment was not associated with a reduced risk of cardiovascular or all-cause hospitalizations (Table 3). Duration >6 months, patients with LVSD, eplerenone, and 50mg subgroups respectively reduced cardiovascular deaths by MRA treatment (RR 0.84, 95%CI (0.75,0.95), P=.006, I2=0%; RR 0.85, 95%CI (0.75,0.96), P=.007, I2=26%;; RR 0.85, 95%CI (0.75,0.96), P=.008, I2=0%; RR 0.85, 95%CI (0.75,0.96), P=.007, I2=0%, Online Resource 4).
3.5. Recurrent myocardial infarction and ventricular arrhythmias
The recurrent myocardial infarction rate was 5.15% (N=257/4992) in those treated with MRA compared with 5.33% (N=266/4988) in control group. The incidence of ventricular arrhythmias was 2.11% (N=99/4702) under MRA treatment and 2.26% (N=106/4695) in control group. Neither of the statistical estimates reached significance (Table 3).
3.6. Changes of cardiac structure and function
MRA use improved LVEF with highly heterogeneous results (Table 3). In addition, improvement in left ventricular end-diastolic volume index (LVEDVI) and end-systolic volume index (LVESVI) was also apparent (Table 3), and further analysis demonstrated a reduction in left ventricula end-diastolic daimeter but not in left ventricula end-systolic daimeter under MRA treatment (Table 3). The ratio of early to late diastolic transmitral flow (E/A ratio) was improved by MRA treatment (Table 3). For LVEF, LVESVI, and LVEDVI, in patients without LVSD (MD 2.74, 95%CI (2.49,2.99), P<.00001, I2=0%; MD −6.23, 95%CI (−10.93,−1.52), P=.009, I2=98%; MD −3.13, 95%CI (−5.79,−0.47), P=.02, I2=29%, Fig. 6), treated ≤6 months (MD 3.86, 95%CI (1.43,6.29), P=.002, I2=93%; MD −5.39, 95%CI (−9.73,−1.04), P=.02, I2=97%; MD −3.41, 95%CI (−5.50,−1.32), P=.001, I2=0%, Fig. 7) and 25mg (MD 2.74, 95%CI (2.49,2.99), P<.00001, I2=0%; MD −6.23, 95%CI (−10.93,−1.52), P=.009, I2=98%; MD −3.13, 95%CI (−5.79,−0.47), P=.02, I2=29%, Online Resource 5) subgroups, the statistical results were significant, respectively. Canrenoate showed the greatest improvement in LVEDVI and LVESVI (MD -4.60, 95%CI (-7.33,-1.86), P=.001, I2=0%; MD −4.46, 95%CI (−7.46,−1.45), P=.004, I2=35%, Online Resource 5). Meanwhile, Spironolactone significantly improved LVEF (MD 3.92, 95%CI (0.71,7.14), P=.02, I2=95%, Online Resource 5).
3.7. Safety
A higher rate of hyperkalemia was 4.79% in the MRA arms versus 2.80% in control groups. Gynecomastia occurred in experiment (0.64%) and control (0.30%) patients. Their overall incidence was nearly 2-fold higher than control groups (RR 1.73, 95%CI (1.44,2.08), P<.00001, I2=42%, Fig. 8). Meanwhile, MRA use increased serum potassium and creatinine levels (MD 0.07 (mmol/l), 95%CI (0.02,0.12), P=.004, I2=74%; MD 0.02 (mg/dl), 95%CI (−0.00,0.04), P=.05, I2=73%, Fig. 8), but no corresponding increase in the incidence of renal dysfunction was found (Table 3). In contrast, hypokalemia occured less frequently in MRA groups (Table 3). Eplerenone (RR 1.48, 95%CI (1.21,1.82), P=.0002, I2=0%), canrenate (RR 3.47, 95%CI (1.43,8.42), P=.006, I2=29%), spironolactone (RR 10.33, 95%CI (2.85,37.41), P=.0004, I2=0%), 25mg (RR 4.91, 95%CI (2.48,9.70), P<.00001, I2=52%), and 50mg (RR 1.48, 95%CI (1.21,1.82), P=.0002, I2=0%, Online Resource 6) subgroups respectively increased the incidence of hyperkalemia. The incidence of gynecomastia was 1.37% in trials using spironolactone and 1.10% in trials using 25mg (RR 8.26, 95%CI (2.23,30.53), P=.002, I2=0%; RR 6.57, 95%CI (1.50,28.83), P=.01, I2=0%, Online Resource 6).
Table 3 other outcomes of mineralocorticord receptor antagonists use in post-AMI patients
|
Heterogeneity
|
Outcomes
|
Trials
|
N
|
RR/MD
|
95%CI
|
P value
|
I2 (%)
|
P value
|
All-cause hospitalizations
|
3
|
6760
|
0.97
|
(0.92,1.03)
|
0.31
|
29
|
0.24
|
Cardiovascular hospitalizations
|
3
|
7690
|
0.92
|
(0.83,1.02)
|
0.10
|
31
|
0.23
|
Recurrent myocardial infarction
|
5
|
9980
|
0.97
|
(0.82,1.14)
|
0.68
|
0
|
0.60
|
Ventricular arrhythmias
|
4
|
9397
|
0.93
|
(0.71,1.22)
|
0.61
|
0
|
0.58
|
Left ventricular ejection fraction
|
8
|
1707
|
2.96
|
(0.96,4.96)
|
0.004
|
92
|
<.00001
|
Left ventricula end-systolic daimeter (cm)
|
3
|
748
|
-0.19
|
(-0.53,0.15)
|
0.26
|
94
|
<.00001
|
Left ventricula end-diastolic daimeter (cm)
|
3
|
748
|
-0.13
|
(-0.26,-0.01)
|
0.04
|
64
|
0.06
|
left ventricular end-diastolic volume index (ml/m2)
|
5
|
1046
|
-3.35
|
(-5.37,-1.34)
|
0.001
|
0
|
0.58
|
left ventricular end-systolic volume index (ml/m2)
|
5
|
1070
|
-4.73
|
(-8.75,-0.70)
|
0.02
|
96
|
<.00001
|
The ratio of early to late diastolic transmitral flow
|
3
|
907
|
0.12
|
(0.10,0.14)
|
<.00001
|
0
|
0,80
|
Renal dysfunction
|
4
|
1534
|
0.45
|
(0.03,6.63)
|
0.56
|
71
|
0.03
|
Hypokalemia
|
3
|
7702
|
0.42
|
(0.19,0.95)
|
0.04
|
64
|
0.06
|
N=Number; MD=Mean difference; RR=Relative ratio; CI=Confidence interval; I2=Inconsistency index