Balancing the risks of bleeding and stent thrombosis: a decision analytic model to compare durations of dual antiplatelet therapy after drug-eluting stents

Abstract

Background: After coronary stent placement, whether dual antiplatelet therapy (DAPT) duration should be extended to prevent late stent thrombosis (ST) or adverse cardiovascular events is uncertain.

Methods: To define the reduction in ischemic events required to outweigh increased bleeding with longer-duration DAPT, we developed a decision-analytic Markov model comparing DAPT durations of 6, 12, and 30 months after DES. Separate models were developed for patients presenting with and without an acute coronary syndrome (ACS). We used sensitivity analyses to identify the incremental benefit of longer-duration DAPT on either ST or the composite of cardiac death, myocardial infarction, and ischemic stroke (major adverse cardiovascular and cerebrovascular events [MACCEs]) required to outweigh the increased risk of bleeding associated with longer DAPT. The outcome from each strategy was quantified in terms of quality-adjusted life years.

Results: In the non-ACS population, in order for 30 months of DAPT to be preferred over 12 months of therapy, DAPT would have to result in a 78% reduction in the risk of ST (relative risk [RR] 0.22, 3.1 fewer events per 1000) and only a 5% reduction in MACCE (RR 0.95, 2.2 fewer events per 1000) as compared with aspirin alone. For the ACS population, DAPT would have to result in a 44% reduction in the risk of ST (RR 0.56, 3.4 fewer events per 1000) but only a 2% reduction in MACCE (RR 0.98, 2.3 fewer events per 1000) as compared with aspirin alone, for 30 months of DAPT to be preferred for 12 months.

Conclusions: Small absolute differences in the risk of ischemic events with longer DAPT would be sufficient to outweigh the known bleeding risks.

Figure 1

Figure 1A: Decision tree representing the outcomes in the Markov model during a given 6-month cycle. Figure 1B State-transition diagram demonstrating transitions between health states in the Markov model. The bubble diagram represents the transitions between the health states in the model. The Markov health states are represented in the ovals with arrows indicating movement between the states from one 6-month period to the next. All patients undergo therapy with DAPT based on the strategy (6, 12, or 30 months) and/or the presence of a prior event requiring further DAPT (e.g. ST or non-fatal MI) based on their original DAPT strategy. Each of the health states were further subdivided on the basis of presence or absence of a prior major bleeding event, but this is not shown for simplicity and brevity. CVA = cerebrovascular accident, DAPT = dual anti-platelet therapy, MI = myocardial infarction, PCI = percutaneous intervention, ST = stent thrombosis

Figure 1

Figure 1A: Decision tree representing the outcomes in the Markov model during a given 6-month cycle. Figure 1B State-transition diagram demonstrating transitions between health states in the Markov model. The bubble diagram represents the transitions between the health states in the model. The Markov health states are represented in the ovals with arrows indicating movement between the states from one 6-month period to the next. All patients undergo therapy with DAPT based on the strategy (6, 12, or 30 months) and/or the presence of a prior event requiring further DAPT (e.g. ST or non-fatal MI) based on their original DAPT strategy. Each of the health states were further subdivided on the basis of presence or absence of a prior major bleeding event, but this is not shown for simplicity and brevity. CVA = cerebrovascular accident, DAPT = dual anti-platelet therapy, MI = myocardial infarction, PCI = percutaneous intervention, ST = stent thrombosis

Figure 2

(panels A–D) Two-way sensitivity analysis – Threshold relative risk (RR) of stent thrombosis (A and B) or MACCE (C and D) as a function of the risk of composite bleeding events (fatal-bleed, major and minor bleed, hemorrhagic stroke) on DAPT for DAPT durations of 12 versus 30 months in the non-ACS (A and C) and ACS (B and D) patient populations. The threshold RR of ST on DAPT as compared to the RR of bleeding on DAPT is demonstrated for the non-ACS population (panel A) and ACS population (panel B), while the threshold RR of MACCE on DAPT as compared to the RR of bleeding on DAPT is demonstrated for the non-ACS population (panel C) and the ACS population (panel D). As the risk of bleeding events on DAPT increases, the threshold relative risk of stent thrombosis on DAPT decreases. For a given RR of bleeding for both ST and MACCE, a lesser relative risk reduction in ST is required for the ACS populations as compared to the non ACS population for 30 months of DAPT to be preferred over 12 months of DAPT after PCI. Based on our meta-analysis, the expected relative risk of bleeding on DAPT as compared to aspirin alone was 1.70. At this risk of bleeding, a RR of ST on DAPT of 0.22 in the non-ACS population and 0.56 in the ACS population would be required to make 30 months of DAPT a favored approach (as denoted by the intersection of the vertical dashed line starting at a RR of bleeding of 1.70 and the horizontal line starting from the y-axis). Beyond a relative risk of bleeding on DAPT of 1.92 in the non-ACS population and 2.61 in the ACS population, no reduction in ST is sufficient to outweigh the bleeding risk. The required relative risk of MACCE on DAPT to make 30-months of DAPT preferred over 12-months of DAPT was 0.95 for the non-ACS (panel C) and 0.98 for the ACS (panel D) patient populations. Substantial increases in the risk of bleeding on DAPT had little effect on the threshold RR of MACCE needed to make longer-duration DAPT preferred. ACS = acute coronary syndrome, MACCE = major adverse cardiovascular and cerebrovascular events, RR = relative risk, all other abbreviations same as in prior figure

Figure 3

Figure 3A: Forest plot showing multiple one-way sensitivity analyses (Non-ACS model). Figure 3B: Forest plot showing multiple one-way sensitivity analyses (ACS model). These tornado graphs demonstrates the relationship between plausible variations in key model parameters and their effect on determining whether longer-term DAPT (30 months) is preferred over 12 months therapy (using the threshold relative risk of ST with DAPT, 0.22 in the non-ACS and 0.56 in the ACS population). Each horizontal bar represents the selected variable’s incremental values (30 month strategy minus 12 months) generated by varying the variable in the range represented. The model was most sensitive to variations in probabilities of VLST on aspirin, hemorrhagic CVA on DAPT, risk of death from ST, relative risks of fatal bleed with DAPT and future non-cardiac death with prior major bleed, and utility associated with prior CVA. *The higher range values for these variables results in 30 months strategy being preferred. ^ Increasing values of the estimate for these variables results in 12 months strategy being preferred. CI = confidence interval QALY = quality adjusted life years, all other abbreviations same as in prior figures

Figure 3

Figure 3A: Forest plot showing multiple one-way sensitivity analyses (Non-ACS model). Figure 3B: Forest plot showing multiple one-way sensitivity analyses (ACS model). These tornado graphs demonstrates the relationship between plausible variations in key model parameters and their effect on determining whether longer-term DAPT (30 months) is preferred over 12 months therapy (using the threshold relative risk of ST with DAPT, 0.22 in the non-ACS and 0.56 in the ACS population). Each horizontal bar represents the selected variable’s incremental values (30 month strategy minus 12 months) generated by varying the variable in the range represented. The model was most sensitive to variations in probabilities of VLST on aspirin, hemorrhagic CVA on DAPT, risk of death from ST, relative risks of fatal bleed with DAPT and future non-cardiac death with prior major bleed, and utility associated with prior CVA. *The higher range values for these variables results in 30 months strategy being preferred. ^ Increasing values of the estimate for these variables results in 12 months strategy being preferred. CI = confidence interval QALY = quality adjusted life years, all other abbreviations same as in prior figures