Biomedicines, Vol. 10, Pages 3159: Efficacy and Safety of TiNO-Coated Stents versus Drug-Eluting Stents in Acute Coronary Syndrome: Systematic Literature Review and Meta-Analysis

a

The risk of bias common to all RCTs was the operator’s knowledge of the type of stent. However, given the absence of significant differences in the baseline and procedural data, this risk seems to have had no effect. Other potential risks of bias appeared occasionally in some RCTs, but the sensitivity analysis shows that bias related to individual RCTs has little influence on the pooled outcomes. The study with the highest risk of bias has the smallest relative weight and does not report outcomes in ACS or at the 5-year follow-up, so the impact of the potential risk of that trial is limited.

b

Heterogeneity within this subgroup was low given the overlap of all the confidence intervals, the Cochran Q-test with p > 0.05, and an I² < 40%.

c

The three ACS RCTs report modified device-oriented MACE. Assuming half of the non-fatal MIs were in non-target vessels (i.e., 45 cases), the proportionality would lead to 27 fewer cases with TiNOS and 18 with DES. If these numbers were removed from the count of MACE, the RR would be 0.98 [0.73, 1.30], which does not change the non-inferiority conclusion.

d

The 95% CI of the RR of MACE included 1.0 with a lower bound slightly <0.75 in favour of TiNOS and a length slightly >0.5. This result is robust to the sensitivity analysis. A hypothetical single RCT to demonstrate the non-inferiority of TiNOS compared to DES and incorporating 1:1 randomization, a 2-sided test, α = 2.5%, power = 90%, and a reference rate of events in the three trial DES arms 99/1123 = 0.088 assuming a non-inferiority margin delta = 0.035 based on a targeted 30% RRR, would require a total sample size of n = 2760 patients, which is 1.01 times the pooled sample size. Considering that the total number of events observed is 124 + 99 = 223, reaching the GRADE rule of thumb of 300 events overall requires n = 2760 × 300/223 = 3713 patients to meet the OIS criterion. However, with the current sample size, a minimum of 20 events (9% of observed events) should be redistributed from the DES arm to the TiNOS arm for the CI to reach non-significance (RR 1.42 [1.06, 1.91]).

e

Publications usually do not indicate whether the MI is related to the target vessel or not, but this review assumes that recurrent MIs related to non-target vessels at a 1-year follow-up should be in equal proportion in the two treatment arms.

f

The 95% CI of the RR of ST is significantly <1.0 in favour of TiNOS with an upper-bound = 0.75 and a length < 0.5. This result is robust to sensitivity analysis. A hypothetical single RCT, 1:1 randomization, 2-sided test, α = 5%, power = 80%, and a reference rate of events in the 3 trial DES arms 69/1123 = 0.061 assuming delta = 0.018 based on a targeted 30% RRR requires a total sample size of n = 4788 patients, which is 1.75 times the pooled sample size. Considering the total number of events observed is 48 + 69 = 117, reaching the GRADE rule of thumb of 300 events overall requires n = 4788 × 300/117 = 12,277 patients to meet the OIS criterion. With the current sample size, a minimum of 9 events (7.7% of observed events) should be redistributed from the DES arm to the TiNOS arm for the CI to reach non-significance (RR 0.71 [0.49, 1.02]). This result is imprecise and fragile.

g

TLR defined as clinically driven confirmed in all three ACS trials

h

The 95% CI of the RR of TLR is significantly >1.0 in favour of DES with a lower bound < 1.25 and a length > 1. This result is not robust to sensitivity analysis. A hypothetical single RCT, 1:1 randomization, 2-sided test, α = 5%, power = 80%, and a reference rate of events in the 3 trial DES arms 46/1123 = 0.041 assuming delta = 0.012 based on a targeted 30% RRR requires a total sample size of n = 7308 patients, which is 2.66 times the pooled sample size. Considering the total number of events observed is 98 + 46 = 144, reaching the GRADE rule of thumb of 300 events overall requires n = 7308 × 300/144 = 15,225 patients to meet the OIS criterion. With the current sample size, a minimum of three events (2.1% of observed events) should be redistributed from the TiNOS arm to the DES arm for the CI to reach non-significance (RR 1.40 [1.00, 1.97]). This result is imprecise and fragile.

i

Recurrent MI is described as non-fatal in the three RCTs and the PK confirms that those included non-fatal MIs in non-target vessels. One can reasonably assume the index stent does not affect those events. Ninety non-fatal MIs were reported. Assuming half of them are related to non-target vessels (i.e., 45 cases), the proportionality with the sample size would lead to 27 fewer cases with TiNOS and 18 with DES, which would result in an RR of 0.19 [0.09, 0.42]. The inclusion of non-target MIs thus results in a dilution that is favourable to DES.

j

The 95% CI of the RR of non-fatal MI is significantly <1.0 in favour of TiNOS with an upper bound < 0.75 and a length < 0.5. This result is robust to sensitivity analysis. A hypothetical single RCT, 1:1 randomization, 2-sided test, α = 5%, power = 80%, and a reference rate of events in the 3 trial DES arms 55/1123 = 0.049 assuming delta = 0.015 based on a targeted 30% RRR requires a total sample size of n = 6070 patients, which is 2.21 times the pooled sample size. Considering the total number of events observed is 35 + 55 = 90, reaching the GRADE rule of thumb of 300 events overall requires n = 6070 × 300/90 = 20,233 patients to meet the OIS criterion. With the current sample size, a minimum of eight events (8.9% of observed events) should be redistributed from the DES arm to the TiNOS arm for the CI to reach non-significance (RR 0.68 [0.44, 1.05]). This result is imprecise and fragile.

k

Heterogeneity within that subgroup was high given the absence of overlap of confidence intervals, the significant Cochran Q-test with p = 0.005, and an I² ≥ 80%.

l

The definition of CD was assumed to be similar across the three ACS RCTs. This uniform reporting method across the three trials was confirmed by PK.

m

The 95% CI of the RR of CD is non-significant with a length > 0.5. Sensitivity analysis supports no association. A hypothetical single RCT, 1:1 randomization, 2-sided test, α = 5%, power = 80%, and a reference rate of events in the 3 trial DES arms 16/1123 = 0.014 assuming delta = 0.004 based on a targeted 30% RRR requires a total sample size of n = 21,484 patients, which is 7.83 times the pooled sample size. Considering the total number of events observed is 14 + 16 = 30, reaching the GRADE rule of thumb of 300 events overall requires n = 21,484 × 300/30 = 214,840 patients to meet the OIS criterion. With the current sample size, a minimum of three events (10% of observed events) should be redistributed from the TiNOS arm to the DES arm for the CI to reach significance (RR 0.45 [0.23 0.90]). This result is imprecise and fragile.

n

ARC are criteria used in all studies. Probable or definite stent thrombosis occur at 1-y FU.

o

The 95% CI of the RR of ST is significantly <1.0 in favour of TiNOS with an upper bound > 0.75 and a length > 0.5. This result is robust to sensitivity analysis. A hypothetical single RCT, 1:1 randomization, 2-sided test, α = 5%, power = 80%, and a reference rate of events in the 3 trial DES arms 20/1123 = 0.018 assuming delta = 0.005 based on a targeted 30% RRR requires a total sample size of n = 17,136 patients, which is 6.25 times the pooled sample size. Considering the total number of events observed is 14 + 20 = 34, reaching the GRADE rule of thumb of 300 events overall requires n = 17,136 × 300/34 = 151,200 patients to meet the OIS criterion. With the current sample size, a minimum of event event (2.9% of observed events) should be redistributed from the DES arm to the TiNOS arm for the CI to reach non-significance (RR 0.52 [0.26, 1.03]). This result is imprecise and fragile.

p

Heterogeneity within that subgroup was moderate given the overlap of the confidence intervals, but the Cochran Q-test was significant with p = 0.05 and I² ≥ 60%.

q

The definition of TD was assumed to be similar across the RCTs so the risk indirectness was rated as not serious. The fact that the sensitivity analysis showed that removing BASE-ACS from the pooled analysis led to a significant confidence interval in favour of TiNOS did not appear to be related to differences in TD definition.

r

The 95% CI of the RR of TD is non-significant with a length > 0.5. Sensitivity analysis supports no association. A hypothetical single RCT, 1:1 randomization, 2-sided test, α = 5%, power = 80%, and a reference rate of events in the 3 trial DES arms 29/1123 = 0.026 assuming delta = 0.008 based on a targeted 30% RRR requires a total sample size of n = 11,740 patients, which is 4.28 times the pooled sample size. Considering the total number of events observed is 29 + 29 = 58, reaching the GRADE rule of thumb of 300 events overall requires n = 11,740 × 300/58 = 60,724 patients to meet the OIS criterion. With the current sample size, a minimum of four events (6.9% of observed events) should be redistributed from the TiNOS arm to the DES arm for the CI to reach significance (RR 0.60 [0.37 0.97]). This result is imprecise and fragile.

留言 (0)

沒有登入
gif