Bleeding Prognoses After Coronary Stenting Unreliable

Two scores widely used to determine whether patients are at high risk for bleeding after coronary stenting disagreed in 22% of patients, and one score underestimated the risk, warned investigators comparing the approaches.

Bleeding risk is important to know when deciding how long to prescribe dual antiplatelet therapy after percutaneous coronary interventions and how intensive therapy should be, according to experts. Clinicians aim to balance a therapy’s higher risk for bleeding with its reduced risk for ischemic events.

The study followed 7562 patients for a year after coronary stenting to track the occurrence of bleeding and ischemic events after the procedure.

Data were obtained from national registers of patients in Sweden who underwent the procedure during a 5-year period. Bleeding risk was assessed using the Academic Research Consortium for High Bleeding Risk (ARC-HBR) score and the Predicting Bleeding Complications in Patients Undergoing Stent Implantation and Subsequent Dual Antiplatelet Therapy (PRECISE-DAPT) score. The study results were published online in JACC: Cardiovascular Interventions.

Although the scores generally agreed on which patients were at high risk for bleeding, 5% of patients were classified as being at high risk only by ARC-HBR and 17% were classified as being at high risk only by PRECISE-DAPT.

Tools Disagree

“It says something about the uncertainty in this field that the two risk tools don’t align,” said Peter Ueda, MD, PhD, one of the authors, who is from the Department of Medicine, Karolinska Institutet, Stockholm, Sweden.

“These tools are designed to be easily applicable in practice,” added Carl-Emil Lim, MD, another of the authors, who is from the same Department of Medicine.

Both scores allow clinicians to quickly calculate risk from a set of criteria, he explained. US guidelines recommend using PRECISE-DAPT, and European guidelines recommend using either of the two scores.

But the study’s findings mean that “both tools should be used with great caution,” Lim pointed out.

The study is “very insightful because it lets us understand how our tools are working in a real-world construct,” said Usman Baber, MD, from the University of Oklahoma Health Sciences Center in Oklahoma City, who wrote an accompanying editorial.

“One of the key insights is that definitions matter,” said Baber, who was not involved in the study. The scores have different definitions of high bleeding risk and use different criteria to determine this. ARC-HBR uses a set of major and minor criteria to determine whether the patient is at high bleeding risk. In contrast, PRECISE-DAPT calculates a risk score out of 100 using five variables, and a score ≥ 25 indicates high risk.

“If you look at practice guidelines and consensus statements, there isn’t that level of broad consensus that we should focus on just one” score, he said, unlike other areas of cardiology, which have an agreed-upon single risk-assessment tool.

Why Scores Can Be Incorrect

The study showed that PRECISE-DAPT underestimated the risk for bleeding across almost all scores. ARC-HBR is not as specific, yielding a simple yes or no answer to whether the patient is at high risk. “The claims that they make are quite different. The bar is higher for PRECISE-DAPT to say it does what it aims to do,” said Lim.

A major issue is that the risk estimated by a score must correspond to the risk observed in the patient population, called “calibration.” In this study, PRECISE-DAPT misestimated the risk, Lim and Ueda reported.

There are various reasons that a score based on many published studies would not have the same predictive value in all populations. Although these scores are based on criteria shown to affect bleeding risk, other factors, such as age and socioeconomic status, may not be considered, Lim said.

And “there are many other determinants of bleeding risk, even if you standardize for risk-factor prevalence,” Ueda said.

Populations are changing, noted Baber, with more patients who are older and have other conditions including cancer.

This creates a dilemma for clinicians. “The effort to predict bleeding remains an important exercise, as it will inform a complex decision concerning antiplatelet therapy,” he said. But “our ability to predict is modest at best.”

Clinicians must assume that bleeding risk estimates derived from risk tools correspond to the actual risk for the patient, and that the risk for ischemic events is similar to that used to develop the score, according to Lim and Ueda.

But the assumptions underlying the current tools are “vague and fuzzy,” Ueda noted.

The choice of which risk-assessment tool to use is important not only when treating patients but also when conducting trials and making regulatory decisions on drug indications linked to bleeding risk, said Baber.

And a larger problem is that risk scores remain underused. “We have spent a lot of time developing tools and scores, but our capacity to use them remains suboptimal. We need to spend a similar amount of effort implementing these tools in clinical workflows,” he added.

Improving Risk Prediction

Tools that better characterize risk — and efforts to get them into widespread use — are needed, the experts agree, but this poses a challenge.

The development of new tools should be encouraged, but the problem of generalizing them to different populations remains. Ueda and Lim suggested two possible approaches. The first is to develop risk scores that can be recalibrated to the local population, similar to scores used for the primary prevention of cardiovascular disease that apply different risk models in countries with varying rates of disease. Another approach is to develop a tool that takes a relative approach and identifies which patients have a larger increase in the relative risk for bleeding and which have a smaller reduction in the relative risk for ischemic protection from long-term dual antiplatelet therapy.

“The relative risk reduction with extra antiplatelet therapy may differ according to many patient characteristics,” Ueda pointed out, as may the ischemic benefit for a given patient. “And there may be interactions between risk factors and treatment.”

The calculus for antithrombotic therapy has become “more complex and nuanced,” Baber said, given the choices of therapies and drug-eluting stents and individual patient characteristics. The decisions can include how long to give therapy, the intensity of therapy, when to decrease the dose, and which drugs to combine.

He said he recommends assessing both bleeding risk and ischemic risk as part of a tailored approach to each patient, and then choosing an optimal antiplatelet strategy from a growing number of options.

“Bleeding as a determinant of poor outcomes is much better understood than before,” Baber said. “Although our ability to predict bleeding is not great, we still have tools available.”