Anticoagulation in pediatric cancer–associated venous thromboembolism: a subgroup analysis of EINSTEIN-Jr

In adults, venous thromboembolism (VTE) is a frequent complication of cancer, and its treatment and anticoagulation is associated with an increased bleeding risk.^1,2 Management of anticoagulation for cancer-associated VTE is often challenged by the risk of drug interactions due to polypharmacy use and the requirement of treatment interruptions because of chemotherapy-induced thrombocytopenia, adverse events, surgery, and invasive procedures, such as lumbar puncture, bone marrow aspiration, and drain placement.³ In contrast with adults, little is known about anticoagulant management in children with cancer-associated VTE.

Recently, the EINSTEIN-Jr study compared the oral direct factor Xa inhibitor, rivaroxaban, with standard anticoagulants in 500 children of all ages for treatment of acute VTE of any type.⁴ Bodyweight-adjusted pediatric rivaroxaban–dosing regimens using a tablet or oral suspension formulations successfully targeted the adult rivaroxaban exposure range without requiring laboratory monitoring.^5,6 Recurrent VTE occurred infrequently with both rivaroxaban and standard anticoagulants, and no major bleeding event was observed in the 335 children who received rivaroxaban.⁴ The absolute incidences of study outcomes and relative treatment effects observed with rivaroxaban were similar to those seen in the larger rivaroxaban VTE studies in adults.^7-10 

In this subgroup analysis of the EINSTEIN-Jr study, we report on the clinical presentation and clinical outcomes of children with cancer-associated VTE and describe the management of anticoagulation in children who had anticoagulant treatment interruptions because of chemotherapy-induced thrombocytopenia, adverse events, or invasive procedures. Furthermore, we explored the pharmacokinetic (PK) parameters of rivaroxaban.

Using data of the EINSTEIN-Jr phase 3 study (clinicaltrials.gov #NCT02234843),^4,10 which included 500 children with VTE who were treated with either the oral direct factor Xa inhibitor, rivaroxaban, or standard anticoagulants, we describe the clinical presentation, risk of bleeding, and recurrent thromboembolism in the subgroup of children who presented with active cancer–associated VTE. Cancer was defined as active in the presence of metastases or whether it was recently (<6 months) diagnosed or treated. In addition, we describe the anticoagulant management and clinical outcomes in children who developed chemotherapy-induced thrombocytopenia, underwent invasive procedures, or had adverse events.

The protocol was approved by the institutional review board at each participating center. Written permission from a parent or a guardian and when appropriate, child assent, were obtained. This substudy was conducted at 33 sites in 13 countries.

Children with confirmed VTE were considered for study inclusion if they had initiated heparin treatment. The main exclusion criteria were active bleeding or high risk of bleeding contraindicating anticoagulant therapy, a platelet count of <50 × 10⁹/L, an estimated glomerular filtration rate <30 mL/min per 1.73 m², and the concomitant use of strong inhibitors of the cytochrome P450 isoenzyme 3A4 (CYP3A4), and/or P-glycoprotein (P-gp), as well as the concomitant use of strong inducers of CYP3A4. The full list of eligibility criteria is provided elsewhere.⁴ Enrollment started with children aged 12 to 17 years followed by those aged 6 to 11, 2 to 5, and 0.5 to 1 years and younger than 0.5 years.

After the completion of the initial heparin treatment, children were randomized in a 2:1 ratio to an open-label therapeutic dose of oral rivaroxaban (tablets or suspension formulation) or a continuation of therapeutic dose of standard anticoagulants (ie, heparins or vitamin K antagonists). Rivaroxaban was administered in a bodyweight-adjusted 20-mg–equivalent daily dose based on phase 1 and 2 data and comprehensive PK modeling predictions in either a once-daily, twice-daily, or thrice-daily regimen in children with a bodyweight of ≥30 kg, ≥12 to <30 kg, or <12 kg, respectively (Table 1).^5,6,11-13 In children weighing <12 kg, the lower range of the adult rivaroxaban exposure was targeted to avoid excessive concentrations at the end of the dosing interval.

The main treatment duration was 3 months, during which children were followed up for the occurrence of recurrent symptomatic VTE and bleeding. At 3 months, repeat imaging of the VTE was performed depending on feasibility. Detailed information was collected on episodes of thrombocytopenia, invasive interventions, adverse events, and concomitant treatments, including cancer-associated medication. In case of an invasive intervention, the study protocol advised to stop anticoagulation at least 24 hours before the intervention, if possible, and restart after the intervention in therapeutic doses within 24 hours, provided adequate hemostasis had been established. Anticoagulant management in children who developed thrombocytopenia was left to the discretion of the treating physician.

A blinded and independent adjudication committee evaluated all baseline and repeat VTE imaging, bleeding events, and symptomatic recurrent VTEs. Bleeding events were graded as major or clinically relevant non-major (CRNM) bleeding. The committee classified the degree of vein recanalization at 3 months as normalized, improved, no relevant change, or deteriorated.¹⁰ 

Blood samples for rivaroxaban PK were taken within specified time windows and analyzed at a central laboratory, as described elsewhere.⁶ 

A comprehensive pediatric population PK model based on a previous pediatric model version and PK data pooled from all the preceding rivaroxaban pediatric studies^5,11-13 was used to evaluate rivaroxaban PK. The following main rivaroxaban PK parameters were derived at steady state for each individual: area under the plasma concentration–time curve from time 0 to 24 hours (AUC_(0-24)ss) as a measure for daily exposure, maximum plasma concentration (C_max,ss), and concentration at the end of the dosing interval (C_trough,ss). Individual results of children with cancer-associated VTE were plotted as a function of bodyweight and compared with the adult reference range (obtained from 203 adults with VTE, younger than 45 years of age who had received 20 mg of rivaroxaban once daily),¹⁴ and results obtained from 281 children with VTE but without cancer who were treated with rivaroxaban, as previously published.⁶ 

Efficacy outcomes were considered during the 3-month study treatment period, whereas safety outcomes were considered for the same period but only from the administration of the first dose of study medication to 48 hours after the last dose. Because of the low frequency of clinical outcomes, data are primarily presented for the entire cohort. 95% confidence intervals (CIs) for incidences were calculated by exact methods. Calculations were performed using SAS 9.2 (SAS Institute Inc, Cary, NC).

A total of 56 children with cancer-associated VTE were randomized (Figure 1). Demographic, clinical, and radiological characteristics are shown in Table 2. The median follow-up during the study period was 91 days (interquartile range, 85-95). Three children did not take the allocated study medication with rivaroxaban. Thirty-six (64.3%) children had a hematologic malignancy of which 23 (63.9%) had acute lymphoblastic leukemias, whereas 20 (35.7%) children had a variety of 13 solid tumors (35.7%; Table 2). The median number of concomitant medications was 30 (interquartile range, 20-49).

The percentage of children with cancer-associated VTE relative to all 500 children in the entire EINSTEIN-Jr study was 11.2% and was highest for children in the age group of 12 to 17 years (51.8%), declining with age to 26.8%, 19.6%, and 1.8% for children aged 6 to 11 years, 2 to 5 years, and below 2 years, respectively.

Most cases of VTE involved the upper extremity (n = 16), the jugular vein (n = 10), and the cerebral vein and sinuses (n = 9) and were associated with the recent use of a central venous catheter in 31 (55.3%) children.

Symptomatic recurrent VTE and major bleeding occurred in none of the 56 children (0%; 95% CI, 0.0%-6.0%; Table 3), whereas 1 (1.8%; 95% CI, 0.4%-9.6%) child had a CRNM bleeding related to vomiting-induced Mallory-Weiss esophageal mucosal tear. A single child died during the study period due to the progression of the malignancy (Table 3). Of the 52 children with an evaluable repeat imaging test, complete vein recanalization occurred in 20 (38.5%) children, incomplete recanalization in 24 children (46.2%), and no relevant change in 8 children (15.4%). None had any evidence of thrombus progression.

Sixteen (28.6%) children developed chemotherapy-induced thrombocytopenia, of whom 14 had a platelet count <50 × 10⁹/L (mean minimum platelet count, 14 × 10⁹/L [range, 2-30]). In 13 of these children, anticoagulant therapy was interrupted 38 times in total (Table 4). Platelet transfusions were given to 3 children, including the child in whom anticoagulation was continued. In an additional 12 children, anticoagulant therapy was interrupted 29 times in total because of lumbar puncture (n = 23) or other invasive interventions (n = 6). Three children had interruptions in the anticoagulant treatment because of CRNM or minor bleeding (n = 3). Overall, the anticoagulant treatment was interrupted in 26 (46.4%) children 70 times, with a mean individual total duration of interruption of 5.8 days, which varied depending on the reason for interruption (Table 4). Anticoagulant therapy was resumed after all treatment interruptions (n = 70) in therapeutic doses and none of the 26 children developed a recurrent VTE, progression of existing thrombus, or a clinically relevant bleeding complication. Normalization on repeat imaging occurred in 10 of the 26 children (38.5%) who had no treatment interruption vs 10 of the 26 children (38.5%) who had their treatment interrupted (risk ratio, 1.0; 95% CI, 0.5-2.0).

Of the 40 children with cancer-associated VTE in the rivaroxaban arm, 35 children were evaluable for PK analyses. Generally, results were comparable for children with or without cancer-associated VTE (Figure 2). Most of the individual values for AUC_(0-24)ss, C_max,ss, and C_trough,ss were within the 5th to 95th percentile of the adult exposure range (Figure 2). As intended, in children with bodyweight <12 kg, values for AUC_(0-24)ss scattered below the median of the adult exposure range with decreasing bodyweights. In children with bodyweight <7 kg, some AUC_(0-24)ss values even fell below the fifth percentile of the adult exposure range but were still in the range of individual adult values below the fifth percentile. Trough values in children were above the lower threshold of the adult exposure range. None of the children were administered strong inhibitors of CYP3A4 and/or P-gp or strong inducers of CYP3A4. No influence of weak and moderate CYP3A4 inhibitors and/or P-gp inhibitors and CYP3A4 inducers on rivaroxaban clearance was identified.⁶ 

The results of our analyses suggest that anticoagulant therapy with either rivaroxaban or standard anticoagulants in children with cancer-associated VTE is safe, efficacious, and associated with complete or partial vein recanalization in almost 85% of the children. Almost two-thirds of children had a hematologic malignancy with acute lymphoblastic leukemias occurring most frequently. Anticoagulant treatment was interrupted because of thrombocytopenia, invasive procedures, or adverse events in almost half of the children. It was resumed in therapeutic doses and was not associated with bleeding, recurrent VTE events, or asymptomatic thrombus deterioration. Despite a significant polypharmacy use, rivaroxaban exposures were within the adult exposure range and comparable with children with VTE who did not have cancer.

Given the coagulopathy associated with active malignant disease in general and hematologic malignancies in particular,^15,16 the absence of thrombotic complications and the low incidence of clinically relevant bleeding in the entire study cohort and in the relatively large group of children who had interruptions in their anticoagulant treatment is notable. In a systematic review and meta-analysis of more than 5000 adult patients with cancer-associated VTE, 6-month incidences of recurrent VTE, major bleeding, and CRNM bleeding with direct oral anticoagulants were 5.8%, 5.5%, and 12.3%, respectively, compared with low-molecular-weight heparin, which were 8.8%, 3.2%, and 6.7%, respectively.¹⁷ In addition, studies with adults with cancer-associated VTE requiring periprocedural interruption of anticoagulation demonstrated high rates of postprocedure thrombotic and bleeding complications.³ The reasons for the observed differences in incidences of thrombotic and bleeding complications between children and adults with cancer-associated VTE are likely multifactorial, including differences in types of cancer, cancer response rates to treatment, comorbid conditions, developmental hemostasis, and vascular aging. Regardless of the underlying reasons for the differences in complication rates between children and adults with cancer-associated VTE, the data presented here suggest that therapeutic anticoagulation is generally safe and effective in children with cancer-associated VTE.

Strengths of our study include the prospective study design, the central blinded outcome evaluation, availability of repeat imaging in almost all children, indication for anticoagulation according to current guidelines,^18,19 and complete follow-up. Several limitations also warrant comment. First, the EINSTEIN-Jr study was designed as a randomized trial comparing rivaroxaban with standard anticoagulation. However, because of the absence of major clinical outcomes in this substudy with 56 children, we decided to present the current data as a single cohort of children with cancer-associated VTE, as to our knowledge, such reports are not available so far. Therefore, we focused on describing demography, clinical presentation, the effect of anticoagulation on thrombus burden, and management of anticoagulation at the time of thrombocytopenia, invasive procedures, and adverse events, and refrained from formal statistical comparisons between rivaroxaban and standard anticoagulation. Hence, our subgroup analysis cannot infer the efficacy or safety of rivaroxaban compared with standard anticoagulation but rather anticoagulation as a whole. Second, our study documented efficacy and safety outcomes limited to a 3-month study period, but anticoagulation could be given for longer durations. Indeed, more than half of the inception cohort was treated with anticoagulant therapy beyond the study period. However, in patients with cancer-associated VTE, thrombotic and bleeding outcomes occur most frequently during the first 3 months.¹⁷ Third, our data do not support a definitive rule as to when to stop anticoagulation in the setting of thrombocytopenia, a reasonable approach would be to follow the same principles used for low-molecular-weight heparin in adults,²⁰ which is to cease or reduce dose administration when the platelet count is <50 × 10⁹/L and restart once platelet recovery has occurred. Unfortunately, based on individual circumstances, clinicians will have to decide whether platelet transfusion and ongoing anticoagulation or cessation or dose reduction of anticoagulation is appropriate.

In conclusion, anticoagulation with either rivaroxaban or heparins with or without vitamin K antagonists appeared safe and efficacious and were associated with reduced clot burden in most children with cancer-associated VTE, including the high number of children who had interruptions in their anticoagulant treatment because of chemotherapy-induced thrombocytopenia, invasive procedures, or adverse events. Rivaroxaban exposures were within the adult exposure range despite substantial polypharmacy use.

The authors thank the children who participated in this study and their supportive families, as well as the investigators, subinvestigators, and coordinators at each of the study sites. The authors would like to thank Chris Woods for his graphical artistic assistance.

This work was funded by Bayer AG and Janssen Research & Development, LLC.

Contribution: J.S.P., A.W.A.L., P.M., and C.M. designed the study; and all authors contributed to data collection, data analysis, data interpretation, writing of the manuscript, approval of the final version, and agreed to be accountable for all aspects of the report.

Conflict-of-interest disclosure: A.W.A.L., A.F.P., M.M., D.K., S.W., and K.T. are the employees of Bayer. C.M. reports personal fees and fees paid to his institution from Anthos, Bayer, Bristol-Myers Squibb, Janssen, Norgine, Pfizer, and Boehringer Ingelheim. H.v.O. reports fees paid to her institution from Bayer, Boehringer Ingelheim, and Octopharma. G.K. reports personal fees and fees paid to her institution from Bayer and Pfizer. The remaining authors declare no competing financial interests.

Correspondence: Joe Palumbo, Department of Pediatrics, Cancer and Blood Diseases Institute, Cincinnati Children's Hospital Medical Center, University of Cincinnati College of Medicine, 3333 Burnet Ave, Cincinnati, OH 45229; e-mail: joe.palumbo@cchmc.org.