But 10-item score may need better accuracy before heading to clinic
This article is a collaboration between MedPage Today® and:
The predictive model, called the S2TOP-BLEED score, quantified the association between patient variables and major bleeding in randomized clinical trials to estimate the 3-year major bleeding risk in people using antiplatelet agents for secondary stroke prevention. It had a concordance statistic (C-statistic) of 0.63 (95% CI 0.60-0.64), reported Nina Hilkens, MD, of the University Medical Center Utrecht in the Netherlands, and colleagues in Neurology.
"Based on 10 readily available characteristics, doctors can estimate the risk of a major bleed on antiplatelet therapy for an individual patient," Hilkens told MedPage Today. "In patients with high risk of bleeding, doctors may want to prescribe additional gastro-protective drugs and monitor blood pressure more closely."
The team performed a meta-analysis of individual patient data from published trials completed prior to December 2010 that looked at antiplatelet therapy in long-term secondary prevention after a TIA or stroke. They selected only trials that randomized patients to aspirin or to other antiplatelet drugs that generally are recommended as first-line treatment to prevent secondary strokes.
Six randomized clinical trials -- CAPRIE, 5 ESPS-2, MATCH, CHARISMA, ESPRIT and PRoFESS -- met their criteria. The researchers excluded patients with strokes of possible cardioembolic origin (such as patients with a history of atrial fibrillation) and patients who were randomized to dipyridamole (Persatine) alone.
The final cohort included 43,112 patients with a TIA or ischemic stroke followed for median time periods ranging from 1.4 to 3.5 years. Of those patients, 1,530 had a major bleeding event, which researchers defined as bleeding within the skull or that resulted in death, hospital admission, or substantial disability. Over 3 years, the risk of major bleeding was 4.6%.
By quantifying the association between patient variables in the trials and the incidence of major bleeding, the researchers found 10 factors that helped predict the risk of bleeding:
• Male Sex
• Type of antiplatelet therapy -- aspirin alone, aspirin/dipyridamole combination agent (Aggrenox), or aspirin with clopidogrel (Plavix)
• High stroke disability score (Outcome on the modified Rankin scale ≥ 3)
• Prior stroke
• High Blood pressure
• Lower BMI
• Asian Ethnicity
History of heart failure was not included because it had varying definitions and conflicting results in the trials.
The researchers produced a calibration plot that showed the correspondence between the predicted and observed risks of all variables and developed a score chart to approximate predictions for individual patients. Of the 43,112 patients, 23,678 patients fell into low-risk categories on the chart, 16,621 were medium risk, and 2,813 were high risk.
Age was the strongest predictor for major bleeding risk. The risk of bleeding ranged from 2% for people 45 to 55 years old with no additional risk factors to more than 10% for people aged 75 to 85 with multiple risk factors.
This is especially important, the authors noted, because about 30% of strokes occur in patients over age 80. "As the population ages, more elderly patients will take antiplatelet drugs and the proportion of people likely to experience a serious bleed increases," Hilkens said.
The researchers validated their model externally using the PERFORM trial, a study of 18,417 patients with a recent TIA or ischemic stroke who were randomized to the novel agent terutroban or aspirin. The external validation showed that S2TOP-BLEED had a C-statistic of 0.61 (95% CI 0.59-0.63) and slightly underestimated the risk of major bleeding.
In an accompanying editorial, Robin Lemmens, MD, PhD, of University Hospitals Leuven in Belgium, and Rustam Al-Shahi Salman, PhD, of the University of Edinburgh, wrote that the S2TOP-BLEED score might help clinicians when they are choosing anti-platelet strategies, but its accuracy needs to improve before it can be used for individual patient decisions.
One way to do that, they noted, could be to incorporate markers of cerebral small vessel disease into the score. "Because of the well-recognized association between the presence of microbleeds on MRI and intracranial bleeding, future iterations of the S2TOP-BLEED score should assess the predictive value of microbleeds, perhaps using data from randomized trials with MR imaging," they wrote.
Limitations of the study include the fact that the prediction model was built with data from trial participants, who might not reflect the entire stroke population. Because patients with a history of bleeding were excluded, the study might underestimate bleeding risk. Trials published before 2010 were used in this study, and stroke diagnosis and treatment have changed since then.
Although the S2TOP-BLEED score may help identify patients at high risk of major bleeding, it does not aim to guide treatment choices, the authors noted, cautioning that the risk of bleeding always should be balanced against the risk of recurring ischemic events.
Hilkens reported no disclosures relevant to the manuscript. Other study authors report relationships with Boehringer Ingelheim, Sanofi, Amgen, Allergan, AstraZeneca, Bayer, BMS, CoAxia, Covidien, Daiichi Sankyo, D-Pharm, GlaxoSmithKline, Johnson & Johnson, Lilly, MSD, Medtronic, MindFrame, Neurobiological Technologies, Novartis, Pfizer, Servier, St. Jude, FibroGen and WebMD Global.
Editorialists Lemmens and Salman reported no disclosures relevant to the manuscript.
- Reviewed by Henry A. Solomon, MD, FACP, FACC Clinical Associate Professor, Weill Cornell Medical College and Dorothy Caputo, MA, BSN, RN, Nurse Planner
NeurologySource Reference: Hilkens N, et al "Predicting major bleeding in patients with noncardioembolic stroke on antiplatelets." Neurology 2017; DOI: 10.1212/WNL.0000000000004289.
NeurologySource Reference: Lemmens R and Salman R, "Individualized risk prediction of major bleeding in secondary stroke prevention." Neurology 2017; DOI: 10.1212/WNL.0000000000004303.