HCT-VTE-J Model to Predict Late-Onset Venous Thromboembolism after Allogeneic Hematopoietic Stem Cell...

With improved survival after allogeneic hematopoietic stem cell transplantation (allo-HSCT), late complications, including venous thromboembolism (VTE), have gained clinical importance. However, the incidence and predictive models for late-onset VTE after allo-HSCT are poorly defined, with limited data on subsequent treatment and outcomes. To describe the clinical features and outcomes of late-onset VTE and establish a risk prediction model for Japanese patients. This single-center retrospective study included 1292 patients (aged ≥16 yr) who underwent their first allo-HSCT between January 2006 and December 2022 at our institution. The primary endpoint was cumulative incidence of VTE, defined as pulmonary thromboembolism (PTE), proximal deep vein thrombosis (DVT), or isolated distal DVT, with catheter-related DVT excluded. Death without VTE is a competing risk factor. Independent risk factors were identified using multivariate Fine-Gray regression. Graft-versus-host disease (GVHD) onset and prolonged hospitalization were treated as time-dependent variables. A risk score (HCT-VTE-J model) was developed based on hazard ratios (HR). Risk discrimination was assessed via the time-dependent area under the receiver operating characteristic curve (AUC). Twenty-eight patients (2.2%) developed VTE at a median of 1226 days after allo-HSCT. Among patients with VTE, the median age and body mass index at the time of allo-HSCT were 54 yr (interquartile range [IQR], 49 to 61) and 23.0 kg/m² (IQR, 21.0 to 25.2), respectively. The median post-transplant hospitalization duration was 73 days (IQR, 58 to 107). All PTE cases (n = 12) were hemodynamically non-massive. At onset, 58.3% (n = 7) of patients experienced extensive chronic GVHD. The initial anticoagulation consisted of heparin (50.0%, n = 6) or dalteparin (41.6%, n = 5). Among DVT cases (n = 16), edoxaban was the most frequently used treatment (50%, n = 8). Thrombus resolution rates were 75.0% (n = 9) for PTE and 56.3% (n = 9) for DVT. Two patients with PTE experienced recurrence, whereas no VTE-related mortality was observed. Multivariable analysis for the incidence of VTE identified five risk factors: age at allo-HSCT ≥50 yr (HR 2.83; 95% confidence interval [CI], 1.24 to 6.45; P = .013), BMI at allo-HSCT ≥22 kg/m² (HR, 2.28; 95% CI, 1.08 to 4.87; P = .031), grade III to IV acute GVHD (HR, 2.62; 95% CI, 1.00 to 6.86; P = .049), extensive chronic GVHD (HR, 4.49; 95% CI, 1.70 to 11.83; P = .002), and post-transplant hospitalization ≥60 days (HR, 3.40; 95% CI, 1.36 to 8.53; P = .009). Based on these factors, the HCT-VTE-J model was constructed, assigning 2 points to extensive chronic GVHD and 1 to each of the remaining four factors. Patients were divided into three groups: low (0 to 1 points), moderate (2 to 3 points), and high-risk (4 to 6 points), with 10-yr incidences of VTE of 0.5% (95% CI, 0.0% to 2.5%), 3.1% (95% CI, 1.3% to 6.3%), and 14.6% (95% CI, 6.5% to 25.8%) (AUC 0.81). We characterized the unique clinical presentation and treatment outcomes of VTE in survivors of allo-HSCT. The developed HCT-VTE-J model enables individualized risk stratification, highlighting the importance of GVHD. Although external validation is warranted, these findings may inform targeted surveillance and thromboprophylaxis strategies.