Individualized Modeling Approach for DAPT Duration

Duration of dual antiplatelet therapy (DAPT) after coronary stenting remains controversial, as do the drivers of the decision-making. Is it the stent, the coronary anatomy, the patient comorbidities, or the reason that led to the coronary intervention? A similar debate took place 10 years ago when acute stent thrombosis was a critical issue believed to be related to high on-treatment platelet reactivity in clopidogrel-treated patients. Scores were developed, some including platelet function tests or genetic variants to guide therapy. Although globally accurate and robust, the relevance of these scores for an individual patient was limited because predicted events are rare. In addition, further randomized evaluations of treatment guidance based on platelet function testing failed to improve clinical outcomes. Finally, a local trigger of stent thrombosis, not visible on angiography, may be present and not considered in the clinical scores, outlining the limited value of these scores in accounting for important confounders.

The issue with prolonged DAPT after drug-eluting stenting is that there is no benefit to expect on mortality except maybe in patients with a previous history of myocardial infarction. The detrimental effect of prolonged DAPT on long-term bleeding is a major determinant of the risk/benefit ratio we can expect from this strategy. The question arises how to interpret this evidence when making decisions for individual patients. There are subgroups of patients in whom the bleeding risk of prolonged DAPT is outweighed by the benefit of reducing stent thrombosis and spontaneous myocardial infarction. Among the difficulties, the weight of the risk factors differs, and some factors expose patients to a higher risk for both an ischemic event and major bleeding. When the knowledge is present, the vision for better care may come with algorithms, predictive scores, or recommendations. Unfortunately, none of these is the perfect solution.

Individualized modeling to identify such patients has been developed by the investigators of the randomized Dual Antiplatelet Therapy Study (DAPT Study) using a net clinical benefit approach, assuming that bleeding and thrombosis confer a comparable weight on subsequent risk. They ended up with a single unified user-friendly tool that was validated in a large independent cohort of stable patients with coronary artery disease. As expected, the model had a modest discrimination and was not powered to examine differences in individual outcomes between subgroups, although it was replicated. In addition, coronary complexity, vessel size, number and length of stents, stent undersizing or malapposition, and dissection, which all are important modulators of the ischemic risks, were not taken into account, whereas many factors were associated with both ischemic and bleeding propensity. However, it is easy to use, and there is a mobile application for bedside decisions.

In this issue of the Journal, Baber et al. report on various clinical and a few procedural risk factors associated with coronary thrombotic events and major bleeding events among patients receiving DAPT after percutaneous coronary intervention for stable coronary artery disease or acute coronary syndrome in the PARIS (patterns of non-adherence to anti-platelet regimens in stented patients) registry. Assigned risk estimates were converted into integer risk scores, allowing categorization into low, intermediate, and high risk for thrombotic and bleeding events, with 2 separate scales. Both scores were validated in the ADAPT-DES (Assessment of Dual AntiPlatelet Therapy with Drug-Eluting Stents) registry. As expected, model performance was moderate, and of note, there was no interaction with stent generation and DAPT cessation. Important baseline characteristics such as body mass index and use of triple therapy are considered in the score, whereas the underlying coronary anatomy and procedural features are not.

The remaining question is whether individualized modeling should be used in routine practice. Let us take a simple example of a 70-year-old man treated a year ago with a second-generation DES (3 mm × 20 mm) implanted at the ostium of the left descending artery for ST-segment elevation myocardial infarction. He has single-vessel coronary artery disease and the left ventricular ejection fraction is 55%. He has a body mass index of 27 kg/m2 and a creatinine clearance of 50 ml/min. He is symptom free. He has a low DAPT score (“1”) suggesting no benefit of extended P2Y12 inhibition, whereas the PARIS score indicates a high thrombotic risk (6 on a scale of 12) and an intermediate bleeding risk score (5 on a scale of 15). This patient could have been included in the DAPT study or in the PARIS registry as well. After the information obtained from the DAPT score, you would probably stop the P2Y12 inhibitor at 1 year, although you would continue on the basis of the PARIS score. Of interest, these scores would not have been different if the patient had 3-vessel disease and a residual SYNTAX score of 8 or more.

There is a general consensus to say that the decision for DAPT prolongation beyond 1 year should be individualized, meaning a case-by-case decision. Whether a strategy based on a score calculation is more efficient than a strategy based on a simple algorithm, expert recommendations, or even just physician appreciation is not known. Scores never reflect the entire truth and interscore variability exists. To overcome confusion, the performance of the various scores should be evaluated prospectively versus an intuitive physician decision.