Patients

We implemented an innovative process for patient identification, including electronic medical record interrogation for eligible patients based upon historical laboratory values suggestive of APS receiving concomitant anticoagulation and direct patient mailing (Figure 1).¹⁵ We enrolled and randomized consenting patients with a history of TAPS receiving therapeutic anticoagulation for secondary prevention of thrombosis for at least 6 months. We classified patients as having definite, likely, or historical APS and gathered additional baseline demographics and clinical characteristics to understand the heterogeneity of APS patients that might inform future research. Definite APS was defined as radiologically verified thrombosis plus serial qualifying laboratory results by Sapporo criteria.¹ Likely APS was defined as radiologically verified thrombosis plus at least 1 qualifying laboratory result, and historical APS was defined as a report of thrombosis along with patient-reported abnormal laboratory testing, but results were not available for confirmation. Patients were excluded if they (1) required anticoagulation for another nonapproved indication; (2) received dual antiplatelet therapy or aspirin >165 mg daily; (3) were pregnant (or with an intention to become pregnant); (4) had a life expectancy of <1 year; (5) had baseline hemoglobin <8 g/dL, platelets <50 × 10⁹/L, creatinine >2.5 mg/dL, total bilirubin >1.5 times the upper limit of normal; or (6) had developed thrombosis while on warfarin with an INR ≥2.0. Randomization occurred using a computer-generated randomization tool with built-in random sequence generation that was applied with allocation sequence concealment without exception.

Open in a separate window

Figure 1.

Consort diagram of patients screened. One MRI screen fail was because of the identification of a brain tumor, and the other was because the patient had white matter changes disproportionate for age. MRI, magnetic resonance imaging.

Interventions

Patients randomized to apixaban received open-label apixaban provided by the study coordinator dispensed at the time of consent and randomization and at follow-up visits. Patients randomized to open-label warfarin were managed per clinical routine. All patients were instructed to contact study personnel immediately with any symptoms worrisome for thrombosis or bleeding, and these signs and symptoms were also reviewed at every encounter.

Follow-up

Following enrollment, patient encounters included in-person visits at months 6 and 12 and remote visits via e-mail portal link or telephone interview with standardized study questionnaires at months 1, 3, 9, and 13. At each visit, adherence to warfarin and apixaban were assessed by standard questionnaires and laboratory tests. Patient satisfaction with anticoagulation treatment was assessed at 1, 3, 6, 9, and 12 months with the Anti-Clot Treatment Scale (ACTS) survey,¹⁶ a 15-item patient-reported survey including 12 items assessing for burden and 3 items assessing for benefit. At the end of the study, all patients randomized to apixaban were transitioned to warfarin using a standardized protocol unless they elected to continue apixaban per routine clinical care at the discretion of the treating physician.

Outcomes

The primary clinical outcome was the combined rate of thrombosis (arterial and venous thrombosis) and vascular death. Arterial thrombosis included ischemic stroke, myocardial infarction, transient ischemic attack, or other arterial thromboembolism defined as we formerly reported.¹⁵ VTE was defined as proximal lower extremity deep vein thrombosis (involving popliteal vein and more proximally) and pulmonary embolism defined in a usual fashion.¹⁵ The primary safety outcome was major and clinically relevant nonmajor bleeding (CRNMB) by the International Society on Thrombosis and Haemostasis criteria.^17,18 All primary efficacy and safety outcomes were assessed by a panel of physicians independent from the study with expertise in thrombosis and APS and who were blinded to the treatment arm. Each was provided outcome event definitions from the protocol to adjudicate all events of thrombosis, death, and bleeding. A κ coefficient of agreement among the independent adjudicators was calculated. We reported patients’ satisfaction using a standardized validated assessment tool.¹⁶

Protocol modifications

Following enrollment of the initial 25 patients, a routine DSMB review occurred. The DSMB observed that 3 strokes, all among patients in the apixaban arm, had occurred and recommended that all future patients randomized to receive apixaban, and all patients already enrolled, receive a higher dose of 5 mg twice daily, and this was adopted. In the subsequent 3 months, 5 additional patients were enrolled, and despite the higher apixaban dose, 3 additional strokes occurred in patients randomized to apixaban. At that time, an ad-hoc DSMB assessment was requested by the principal investigator. After reviewing the data, the DSMB recommended: (1) to continue enrollment, (2) that any patient previously enrolled with a history of arterial thrombosis and randomized to apixaban to be returned to warfarin, (3) that only patients without a history of arterial thrombosis to be considered eligible subsequently, and (4) that brain magnetic resonance imaging using a stroke detection protocol be obtained for all subsequent eligible candidates, and that only patients without radiographic evidence of prior stroke or white matter changes disproportionate for patient age can be enrolled. These recommendations were communicated to the IRB, and all recommendations were adopted, protocol modifications were published,¹⁹ and enrollment ensued until April 2019.

Statistical analysis

Based upon former work,^20,21 we estimated a recurrent rate of thrombosis of 1.5% among patients receiving anticoagulation, and using a noninferiority design, we estimated that 4640 subjects would need to be enrolled to have 80% power to report a 1% difference in thrombosis rate. Given the large number of patients (with a rare condition) needed, we aimed to conduct a pilot study first by randomizing 200 patients to apixaban or warfarin to report primary outcome rates that could evaluate feasibility and inform future research. We hypothesized that apixaban would not differ from warfarin for the outcome of thrombosis and vascular death or major bleeding and CRNMB among patients with TAPS receiving therapeutic anticoagulation for secondary prevention of thrombosis. Descriptive demographics of the patient population were calculated for each arm as either mean and standard deviation or number of participants and percentage. Intention-to-treat analysis was used to report the primary efficacy outcome (thrombosis and vascular mortality) and primary safety outcome (major bleeding and CRNMB). Cox proportional-hazards models were used to assess differences in outcomes between the study arms. Person-time and Kaplan-Meier curves were calculated for each outcome separately and for the net clinical benefit upon combining the primary efficacy and safety outcomes. In a sensitivity analysis, the as-treated status of participants was considered for exposure to apixaban 2.5 mg twice daily dosing, apixaban 5.0 mg twice daily dosing, and warfarin. An additional sensitivity analysis was done to omit patients with a history of arterial thromboembolism. Quality of anticoagulation management was assessed by calculating the time in therapeutic range in a usual fashion.²² INR values used in the calculation were collected as part of the study protocol, as well as all laboratory values gathered from routine clinical care. Adherence to apixaban was calculated from the visit questionnaires that interrogated for elective interruption and missed doses and were reported as a percent of doses prescribed.

To analyze findings from the ACTS survey, benefit and burden questions were aggregated separately and then combined for a final score. Higher scores correlated with greater satisfaction. A Mann-Whitney U test was used to compare between study arms. All analyses were conducted with R version 4.0.3.

Primary efficacy and safety outcomes

The Cox proportional-hazards models did not converge; therefore, comparisons of outcomes between study arms or subgroup analyses and their levels of statistical significance are not reported. The focus of the analysis shifted to a person-time analysis.

In the intention-to-treat analysis for the primary efficacy outcome, there were 6 thrombotic events among patients randomized to apixaban, all of which were strokes, at a rate of 318 events per 1000 person-years. No patient randomized to warfarin experienced a thrombotic event (Figure 2). There was 1 major bleeding event in the warfarin arm (vaginal bleeding with an INR = 2.9) and no CRNMB events, which resulted in a rate of 40 per 1000 person-years. No patient randomized to apixaban experienced a major bleed or CRNMB. When combining the primary efficacy (thrombosis and vascular death) and safety (major bleeding and CRNMB) outcomes, the rate of adverse outcomes per 1000 person-years was 318 for apixaban and 40 for warfarin. The characteristics of individual patients that experienced an outcome event are reported in Table 2, with additional details reported in supplemental Table 3. The κ coefficient of agreement among the independent adjudicators for the outcome events was 1.0.

Open in a separate window

Figure 2.

Kaplan-Meier cumulative event rate for thrombosis. The solid red line is for apixaban, and the solid blue line is for warfarin.

The authors would like to acknowledge Drs. J. Brent Muhlestein, Joseph Bledsoe, Robert Hoesch, Paul Asmar, Coleman Herrod, Logan McLean, Mark Dodson, Megan Donohue, Erica Mulciare-Jones, and Paul Johnson; the clinical research coordinator Abigail Bartosic; and Dr. J. Michael Dean and the staff of the University of Utah Center for Clinical & Translational Science.

Investigator-initiated funding from Bristol-Myers-Squibb-Pfizer Alliance was paid to Intermountain Healthcare.