Estimating Bleeding Risk in Patients with Cancer-Associated Thrombosis: Evaluation of Existing Risk Scores and...

Area of Science:

• Hematology
• Oncology
• Clinical Epidemiology

Background:

• Assessing bleeding risk is crucial for managing cancer-associated thrombosis (CAT).
• Current bleeding risk scores lack validation in CAT patients and are not recommended for clinical practice.
• Effective risk stratification is needed to guide treatment decisions and improve patient outcomes.

Purpose of the Study:

• To compare the predictive performance of existing venous thromboembolism (VTE) bleeding risk scores, a pragmatic cancer type classification, and a novel prediction model for clinically relevant bleeding in CAT patients.
• To evaluate the utility of established risk scores and propose improved methods for bleeding risk estimation in this specific population.

Main Methods:

• A posthoc analysis of the Hokusai VTE Cancer study involving 1,046 patients treated for CAT.
• External validation of seven existing bleeding risk scores (ACCP-VTE, HAS-BLED, Hokusai, Kuijer, Martinez, RIETE, VTE-BLEED).
• Comparison with a pragmatic classification based on cancer type and a newly derived competing risk-adjusted prediction model ('CAT-BLEED').

Main Results:

• Existing risk scores demonstrated poor to moderate predictive performance (C-statistics: 0.50-0.57) with poor calibration.
• The pragmatic classification and the 'CAT-BLEED' model showed moderate internal validation performance (C-statistics: 0.61 and 0.63, respectively) with good calibration.
• 149 clinically relevant bleeding events were analyzed within 6 months post-CAT diagnosis.

Conclusions:

• Established bleeding risk scores are inadequate for patients with cancer-associated thrombosis (CAT).
• A pragmatic classification based on cancer type offers marginal improvement in risk prediction.
• The novel 'CAT-BLEED' model shows potential but requires external validation and demonstration of clinical utility with various direct oral anticoagulants (DOACs).