Cancer-associated arterial thromboembolism: real-world burden from a nationwide electronic health record database

Abstract Background Arterial thromboembolism (ATE), including myocardial infarction (MI) and ischemic stroke (iCVD), is increasingly recognized in patients with cancer; however, contemporary population-level data remain limited. Methods We conducted a retrospective cohort study using Epic Cosmos to identify adults with incident cancer (2018 to 2023). ATE events (MI and iCVD) were ascertained using validated inpatient ICD-10-CM billing codes. Cumulative incidence was estimated using competing risk method with death as a competing event. Multivariable Cox proportional hazards models with LASSO penalization were used for variable selection and risk estimation. Systemic therapy (for ATE) and ATE onset (for mortality) were modeled as time-varying covariates. Results Among 1,740,618 patients, the cumulative incidence of ATE was 2.4%, 3.4%, and 4.9% at 1, 2, and 5 years, respectively. Incidence was highest in acute leukemia (5.5%), lung (5.1%), and pancreatic (4.9%), and lowest in prostate (1.5%), uterine (1.5%), breast (0.9%), and testicular cancers (0.5%). Independent factors associated with ATE onset included cancer-specific characteristics (cancer type, metastatic disease, and systemic therapy); pre-existing cardiovascular disease (prior MI, iCVD, heart failure, peripheral vascular disease); cardiovascular risk factors (older age, male sex, diabetes, hypertension, renal disease); and other clinical and laboratory variables. In an analysis adjusted for baseline confounders, time-varying onset of ATE was associated with an approximately 3-fold increase in mortality. Conclusions ATE incidence and risk vary significantly across cancer populations. The development of ATE is associated with adverse prognosis. These findings highlight the importance of adapted cardio-oncology risk assessment.