A target trial emulation to assess the comparative incidences of venous thromboembolism in patients with chemotherapy induced thrombocytopenia treated with romiplostim.

Abstract Introduction Chemotherapy induced thrombocytopenia (CIT) commonly occurs in patients with solid tumors often resulting in dosage reductions or treatment delays, potentially leading to decreased treatment effectiveness. Romiplostim, a thrombopoietin receptor agonist (TPO-RA), has been shown to increase platelet counts and reduce treatment delays in patients with CIT. Studies have suggested a potential increase in venous thromboembolism (VTE) among patients with immune thrombocytopenia receiving TPO-RAs (PMID:31073079). Patients with cancer undergoing chemotherapy are already at an increased risk of VTE raising concern for a potential pro-thrombotic effect of romiplostim. Comparative data on VTE incidence with and without romiplostim in patients with CIT are lacking. Aim To compare the rates of VTE at 6 months in patients with solid tumors and CIT with or without administration of romiplostim. Methods We performed a target trial emulation to compare outcomes among patients with cancer and CIT who received romiplostim, matched with a control population with CIT that did not receive romiplostim. The romiplostim cohort included consecutive patients aged 18 years or older with solid tumors receiving at least one romiplostim dose for CIT (platelets <100×109/L within 30 days before romiplostim) between 4/1/2010 and 6/16/2024. The no-romiplostim control cohort consisted of patients with thrombocytopenia and solid tumors receiving chemotherapy who were exact matched based on cancer diagnosis and same chemotherapy regimen (for which romiplostim was administered). Romiplostim patients were indexed at first romiplostim dose, and days from start of associated chemotherapy regimen to index were calculated. A quasi-index date for each control was calculated by adding this number of days (from their romiplostim pair) to the start of the control's matched chemotherapy regimen. The control cohort was further matched based on degree of thrombocytopenia within 30 days before index. Final 1:1 matching of the control cohort was performed with a propensity score using age, time from cancer diagnosis to index, metastatic status and Charlson comorbidity index (CCI) at diagnosis, and platelet count, hemoglobin, AST, ALT, eGFR closest to index date. The primary outcome of VTE at any site was identified using a natural language processing (NLP) application that uses predetermined keywords to identify potential clinical events. VTE detected by NLP were reviewed by a hematologist. Manual chart review was used to cross-confirm NLP-identified VTE events and to identify arterial thromboembolism (ATE). Records were followed for 6 months or until VTE or death. Cumulative incidence of outcomes was assessed with death as a competing risk. Results There were 233 romiplostim-treated patients and 233 matched controls with CIT who did not receive romiplostim. There was no difference in baseline characteristics between the romiplostim and control groups. Overall, median age was 61.6 years (IQR 49.9-69.0) and 254 (54.5%) were female. Most common cancer types included breast (98, 21.0%), pancreatic (74, 15.9%), colon (63, 13.5%) and upper gastrointestinal (48, 10.3%), and 396 (85.0%) had metastatic cancer. Median time from cancer diagnosis to index was 10.5 months (IQR: 4.37-29.72). Median time from start of the associated chemotherapy to index was 1.97 months (IQR: 1.05-3.06). Median platelet count (×109/L) at index was 59 (IQR: 40-74). A total of 42 (18.0%) patients receiving romiplostim and 38 (16.3%) controls were diagnosed with VTE prior to index, of which 16 (6.9%) and 19 (8.2%) occurred within 3 months before index, respectively. Anticoagulation was present at index in 45 (19%) patients receiving romiplostim and 57 (24%) controls (p=0.218). ATE occurred prior index in 12 (6.4%) romiplostim patients and 10 (4.3%) controls (p=0.411). The 6-month cumulative incidence of VTE in the romiplostim cohort was 10.0% (95% confidence interval [CI] 6.9-15.0%) and 9.1% (95% CI 5.8-13.0%) in the control cohort (hazard ratio [HR] 1.05; 95% CI 0.58-1.88, Grays test P=0.69). Results were similar when stratifying for prior VTE. The 6-month incidence of ATE was 2.2% (95% CI 0.8-4.7%) in the romiplostim cohort and 2.2% (95% CI 0.8-4.8%) in controls (HR 0.92; 95% CI 0.27-3.18). Conclusions In a large cohort of patients with cancer and CIT the administration of romiplostim was not associated with an increased risk of VTE or ATE.