Risk Assessment Models for Predicting Venous Thromboembolism in Patients with Pancreatic Cancer

Area of Science:

• Oncology
• Hematology
• Clinical Epidemiology

Background:

• Venous thromboembolism (VTE) is a significant risk for pancreatic cancer (PC) patients undergoing chemotherapy.
• Limited data exists on the performance of current risk assessment models (RAMs) for predicting VTE in this specific patient group.

Purpose of the Study:

• To conduct a head-to-head comparison of the Khorana, PROTECHT, and ONKOTEV RAMs.
• To evaluate the predictive accuracy and discriminatory performance of these RAMs for VTE in PC patients receiving outpatient chemotherapy.

Main Methods:

• A prospective, multicenter study of 762 newly diagnosed PC patients (BACAP cohort).
• Khorana, PROTECHT, and ONKOTEV scores were calculated at enrollment before chemotherapy.
• VTE events were tracked for 6 months; time-dependent Brier scores and c-indexes were used for performance assessment.

Main Results:

• The cumulative incidence of VTE at 6 months was 16.4%.
• All RAMs showed good calibration (Brier score: 0.14) but poor discrimination (c-indexes ranging from 0.50 to 0.53).
• Sub-distribution hazard ratios indicated minimal differences in VTE risk between intermediate and high-risk strata for all models.

Conclusions:

• The Khorana, PROTECHT, and ONKOTEV scores exhibit inadequate performance for predicting VTE in PC patients on outpatient chemotherapy.
• There is a critical need for developing and validating novel risk assessment tools to guide thromboprophylaxis strategies in this vulnerable population.