The effect of phenytoin on the pharmacokinetics of tirilazad mesylate in healthy male volunteers.

The pharmacokinetic interaction between phenytoin and tirilazad was studied in 12 healthy men who received 200 mg phenytoin orally every 8 hours for 11 doses and 100 mg for the remaining 5 doses in one period of a two-way crossover study. In both periods, 1.5 mg/kg tirilazad mesylate was administered (as 10-minute intravenous infusions) every 6 hours for 21 doses (5 days). Plasma tirilazad mesylate and U-89678 (an active metabolite) were quantified by HPLC. After dose 21, area under the plasma concentration-time curve [AUC(0-6)] for tirilazad mesylate was significantly lower (p = 0.0061) after phenytoin treatment (3029 +/- 982 ng.hr/ml) than after tirilazad mesylate alone (4647 +/- 1562 ng.hr/ml). AUC(0-6) for U-89,678 after dose 21 was reduced from 1485 +/- 1173 ng.hr/ml after tirilazad mesylate alone to 195 +/- 223 ng.hr/ml after phenytoin coadministration. U-89678 normally accumulates during multiple dosing, but mean U-89678 trough concentrations decreased after 24 hours during tirilazad and phenytoin coadministration. No clinically significant interactions of tirilazad mesylate and phenytoin for medical events, vital signs, or laboratory parameters were identified. These results suggest that phenytoin rapidly induces tirilazad mesylate metabolism; it may also induce the metabolism of U-89678 or shunt tirilazad mesylate metabolism through other pathways.

Effect of age and gender on tirilazad pharmacokinetics in humans.

Tirilazad mesylate pharmacokinetics were assessed in 12 young and 12 elderly volunteers (six men and six women per age group). Subjects received single 10-minute intravenous infusions of 1.5 mg/kg and 3.0 mg/kg tirilazad mesylate. Plasma tirilazad mesylate concentrations were determined by HPLC. There were no significant dose effects on clearance, but half-life increased with dose because of assay insensitivity at the lower dose. Mean half-lives were 16.3 +/- 15.5 and 21.4 +/- 12.6 hours for young and elderly subjects, respectively, at the 3.0 mg/kg dose. At the same dose, mean tirilazad mesylate systemic clearance was 0.630 +/- 0.254 and 0.428 +/- 0.090 L/hr/kg, respectively. The decreased clearance in elderly volunteers was primarily attributable to a lower clearance in elderly women relative to young women. The small effect of age on tirilazad clearance is likely to have minimum clinical impact. Tirilazad clearance was approximately 40% higher in young women than in young men. The clinical importance of this observation is unknown.

Pharmacokinetics and pharmacodynamics of oral diazepam: effect of dose, plasma concentration, and time.

Eleven healthy subjects received single oral doses of placebo, 2 mg diazepam, 5 mg diazepam, and 10 mg diazepam in a randomized four-way crossover study. Plasma diazepam levels, the Digit Symbol Substitution Test (DSST), and fraction of total electroencephalographic (EEG) amplitude falling in the sigma plus beta (13 to 31 Hz) frequency range were determined during the 12 hours after drug administration. Peak plasma diazepam concentration and area under the 12-hour curve were proportional to dose; time of peak was independent of dose. Baseline percentage of EEG amplitude falling in the 13 to 31 Hz range averaged 15.7% and did not differ among the four trials. The percentage of EEG amplitude falling in the 13 to 31 Hz range did not change over baseline with placebo or 2 mg diazepam but was increased 1/4 to 2 1/2 hours after 5 mg diazepam, (maximum, +7.3%) and 3/4 to 12 hours after 10 mg diazepam (maximum, +15.2%). The increase in the percentage of EEG amplitude falling in the 13 to 31 Hz range was highly correlated with plasma diazepam concentration. DSST scores for placebo and 2 mg diazepam were nearly identical. DSST decrements with 5 and 10 mg diazepam paralleled and were correlated with the changes in the percentage of EEG amplitude falling in the 13 to 31 Hz range and with plasma diazepam levels. Thus the EEG analysis provides objective quantitation of benzodiazepine central nervous system effects, in turn reflecting plasma levels and other clinical measures.

Comparison of ximelagatran, an oral direct thrombin inhibitor, with enoxaparin for the prevention of venous thromboembolism following total hip replacement. A randomized, double-blind study.

BACKGROUND: Prophylaxis is recommended following total joint replacement because of the high risk of venous thromboembolism (VTE). Postoperative low-molecular-weight heparin (LMWH) reduces the incidence of venographically detected deep vein thrombosis (DVT) to about 10-15% in total hip replacement (THR) patients. Ximelagatran is a novel, oral direct thrombin inhibitor that selectively and competitively inhibits both free and clot-bound thrombin. We compared the efficacy and safety of ximelagatran with those of enoxaparin for the prevention of VTE in patients undergoing THR. METHODS: This was a prospective, randomized, multicenter, double-blind study conducted principally in the USA and Canada. Patients received fixed-dose oral ximelagatran 24 mg bid or subcutaneous enoxaparin 30 mg bid and matched placebo for 7-12 days; both regimens were initiated the morning after surgery. The incidence of VTE (by postoperative day 12) included thrombosis determined by mandatory venography of the leg on which surgery was performed and symptomatic, objectively proven DVT or pulmonary embolism (PE). VTE and bleeding events were interpreted by an independent central adjudication committee for primary analysis. RESULTS: Of the 1838 patients randomized, 1557 had either adequate venography or symptomatic, proven VTE (efficacy population). Overall rate of venography acceptable for evaluation was 85.4%. Overall rates of total VTE were 7.9% (62 of 782 patients) in the ximelagatran group and 4.6% (36 of 775 patients) in the enoxaparin group, with an absolute difference of 3.3% and a 95% confidence interval for the difference of 0.9% to 5.7%. Proximal DVT and/or PE occurred in 3.6% (28 of 782 patients) in the ximelagatran group and 1.2% (nine of 774 patients) in the enoxaparin group. Major bleeding events were observed in 0.8% (seven of 906) of the ximelagatran-treated patients and in 0.9% (eight of 910) of the enoxaparin-treated patients (P > 0.95). Non-inferiority of ximelagatran 24 mg bid based on a prespecified margin of 5% was not met, resulting in superiority of the enoxaparin regimen. CONCLUSIONS: Both ximelagatran and enoxaparin decreased the overall rate of VTE compared with that reported historically. However, in this study, enoxaparin 30 mg bid was more effective than ximelagatran 24 mg bid for prevention of VTE in THR. Oral ximelagatran was used without coagulation monitoring, was well tolerated, and had bleeding rates comparable to those of enoxaparin. Further refinement by testing a higher dose of ximelagatran in the patients undergoing THR is warranted.

Effects of a fish oil concentrate in patients with hypercholesterolemia.

The effects of a fish oil supplement on lipid and lipoprotein levels, platelet function, and vital signs were investigated in 31 hypercholesterolemic patients. Thirteen patients took 5 g of encapsulated fish oil per day and 18 patients took 5 g of encapsulated safflower oil "placebo" per day for 28 days. Diet and exercise patterns were kept as constant as possible during the study. The fish oil group had significant increases in several lipid/lipoprotein values at the end of the treatment, including an increase of total cholesterol of 14% (P = 0.0001), LDL of 16% (P = 0.003), HDL of 13% (P = 0.015) and HDL2 of 36% (P = 0.009). The triglyceride level fell 24%, a nonsignificant change (P = 0.217). The ratios of total cholesterol/HDL and LDL/HDL were increased at the end of fish oil treatment, and returned to baseline 30 days after fish oil was stopped. The placebo group had no significant changes in any of the lipid/lipoprotein values. Neither the fish oil nor the placebo group had significant changes in vital signs or platelet function tests (bleeding time, thromboxane B2, platelet factor 4 and beta-thromboglobulin) during the study. These results suggest that fish oil supplements may have an adverse effect on lipid/lipoprotein values in hypercholesterolemic patients.

Pharmacokinetics of tirilazad and U-89678 in ischemic stroke patients receiving a loading regimen and maintenance regimen of 10 mg/kg/day of tirilazad.

The pharmacokinetics of tirilazad mesylate and its active reduced metabolite, U-89678, were evaluated in ischemic stroke patients receiving 2.5 mg/kg tirilazad every 3 hours for the first 12 hours of dosing followed by 2.5 mg/kg every 6 hours for a total of 22 doses (5 days). Trough and serial samples drawn during the 6 hours after administration of the last dose were analyzed for plasma levels of tirilazad and U-89678 by means of high-performance liquid chromatography. Complete concentration-time profiles were available for 20 patients, including 12 men (mean age, 68.0 years) and 8 women (mean age, 75.0 years). Trough concentrations of tirilazad and U-89678 were consistent with the loading regimen used. The mean area under the concentration-time curve from time 0 to 6 hours (AUC0-6) of tirilazad was 8181 +/- 2398 ng.hr/mL in men and 8135 +/- 3671 ng.hr/mL in women. The mean AUC0-6 of U-89678 was 2761 +/- 1834 ng.hr/mL in men and 1477 +/- 903 ng.hr/mL in women. These results show that gender has a modest effect on the pharmacokinetics of U-89678 but little effect on the pharmacokinetics of tirilazad in elderly ischemic stroke patients. These observations are consistent with previous findings in healthy young and elderly subjects.

Evaluation of the pharmacokinetics and tolerability of tirilazad mesylate, a 21-aminosteroid free radical scavenger: I. Single-dose administration.

The single-dose tolerability and pharmacokinetics of tirilazad mesylate, a 21-aminosteroid free radical scavenger, were assessed in 47 healthy male subjects. Subjects were randomized to receive citrate vehicle (n = 12) or 0.25 mg/kg (n = 9), 0.5 mg/kg (n = 9), 1.0 mg/kg (n = 8), or 2.0 mg/kg (n = 9) tirilazad mesylate by 0.5-hour intravenous infusion. Injection site pain was observed with approximately equal frequency in both vehicle and tirilazad mesylate treatment groups. No statistically significant effects of tirilazad mesylate on blood pressure, heart rate, electrocardiograms, liver enzymes, or renal function were apparent. Tirilazad mesylate did not significantly affect measures of glucocorticoid activity (blood glucose, adrenocorticotropic hormone, cortisol, eosinophil, or lymphocyte levels). Maximal plasma concentrations of tirilazad mesylate increased linearly with dose. Limited assay sensitivity at the lower two doses prevented determination of the dose proportionality of tirilazad area under the curve. The apparent elimination half-life at the higher doses was 3.7 hours. Clearance of tirilazad mesylate approached liver blood flow. Results indicate that intravenous infusions at these doses are well tolerated and devoid of glucocorticoid effects. Tirilazad mesylate appears to be efficiently cleared by the liver, and its pharmacokinetics are apparently linear over the dosage range studied.

Lack of a pharmacokinetic/pharmacodynamic interaction between nimodipine and tirilazad mesylate in healthy volunteers.

The potential interaction between tirilazad mesylate, a membrane lipid peroxidation inhibitor, and nimodipine, a calcium-channel antagonist, was assessed in 12 healthy male volunteers. Subjects received 60 mg nimodipine orally, 2.0 mg/kg tirilazad mesylate as a 10-minute intravenous infusion, and a combination of the two treatments according to a balanced 3-way crossover design. No significant effects of nimodipine on tirilazad mesylate pharmacokinetic parameters were observed (P > .05). Values for tirilazad mesylate clearance (34.9 +/- 8.96 L/hr) and half-life (29 +/- 7.83 hr) were consistent with previous studies. Nimodipine pharmacokinetic parameters exhibited substantial variability, and mean AUC was approximately 25% below the range of previously published values. However, no significant differences in nimodipine pharmacokinetics were observed between treatments. Nimodipine administration increased heart rate slightly without a change in blood pressure, which was not observed after tirilazad administration and was not altered when tirilazad and nimodipine were coadministered. Thus, no significant interaction between tirilazad mesylate and nimodipine is detectable after single-dose administration.

Steady-state pharmacokinetic properties of pramipexole in healthy volunteers.

Pramipexole is a dopamine receptor agonist that has proved effective in the treatment of Parkinson's disease. The pharmacokinetic properties of pramipexole at steady-state concentrations were studied in 16 healthy men and women at four dose levels throughout the range recommended for Parkinson's patients. Plasma and urine samples collected within the four dose intervals were assayed for concentrations of pramipexole, using high-performance liquid chromatography. The total oral clearance for all participants was 419 mL/min. The mean volume of distribution and elimination half-life for all participants was 486 +/- 93.2 L and 12.9 +/- 3.27 hours. Concentrations of pramipexole were proportional to dose, although the drug's pharmacokinetic properties differed between men and women. The area under the concentration-time curve for each dose level was 35% to 43% greater in women, mainly because of a 24% to 27% lower oral clearance. The mean creatinine clearance in men and women was 112 +/- 12.8 mL/ min/1.73 m2 and 80.9 +/- 15.6 mL/min/1.73 m2, respectively. The renal clearance of pramipexole accounts for approximately 80% of oral clearance, and there was a significant correlation between renal and creatinine clearances. The influence of gender could not be distinguished from the influence of age and the resulting reduced creatinine clearance, but the measurement of pharmacokinetic properties produced linear results in both men and women.

Evaluation of the pharmacokinetics and tolerability of tirilazad mesylate, a 21-aminosteroid free radical scavenger: II. Multiple-dose administration.

The multiple dose tolerability and pharmacokinetics of tirilazad mesylate, a 21-aminosteroid free radical scavenger, were assessed in 50 healthy male volunteers. Volunteers were randomized to receive intravenous normal saline placebo (n = 10), citrate vehicle placebo (n = 10), or 0.5 mg/kg/day (n = 6), 1.0 mg/kg/day (n = 6), 2.0 mg/kg/day (n = 6), 4.0 mg/kg/day (n = 6), or 6.0 mg/kg/day (n = 6) tirilazad mesylate in divided doses every 6 hours for 5 days, for a total of 21 doses. Drug was infused over 10 or 30 minutes. All tirilazad mesylate treatment groups and the citrate vehicle group had significantly more frequent and more intense pain at the injection site than did the saline group, but the pain intensity did not require interruption of dosing. Three episodes of clinical thrombophlebitis were observed. No statistically significant effects of tirilazad mesylate on blood pressure, heart rate, electrocardiograms, or renal function were apparent. Moderate and transient increases in serum alanine transaminase were observed in several subjects. In the 6.0 mg/kg/day group, 50% of the subjects exhibited increased alanine transaminase. Tirilazad mesylate did not significantly affect measures of glucocorticoid activity (blood glucose, adrenocorticotropic hormone, cortisol, eosinophil, or lymphocyte levels). Tirilazad mesylate pharmacokinetics were linear over the dosage range studied. Steady state appeared to be achieved by the fifth day of dosing. After the last dose, a mean terminal half-life of 35 hours was observed.(ABSTRACT TRUNCATED AT 250 WORDS)

Hormonal effects on tirilazad clearance in women: assessment of the role of CYP3A.

This study assessed whether the previously reported difference in tirilazad clearance between pre- and postmenopausal women is reversed by hormone replacement and whether this observation can be explained by differences in CYP3A4 activity. Ten healthy women from each group were enrolled: premenopausal (ages 18-35), postmenopausal (ages 50-70), postmenopausal receiving estrogen, and postmenopausal women receiving estrogen and progestin. Volunteers received 0.0145 mg/kg midazolam and 3.0 mg/kg tirilazad mesylate intravenously on separate days. Plasma tirilazad and midazolam were measured by HPLC/dual mass spectrophotometry (MS/MS) assays. Tirilazad clearance was significantly higher in premenopausal women (0.51 +/- 0.09 L/hr/kg) than in postmenopausal groups (0.34 +/- 0.07, 0.32 +/- 0.06, and 0.36 +/- 0.08 L/hr/kg, respectively) (p = 0.0001). Midazolam clearance (0.64 +/- 0.12 L/hr/kg) was significantly higher in premenopausal women compared to postmenopausal groups (0.47 +/- 0.11, 0.49 +/- 0.11, and 0.53 +/- 0.19 L/hr/kg, respectively) (p = 0.037). Tirilazad clearance was weakly correlated with midazolam clearance (r2 = 0.129, p = 0.02). Tirilazad clearance is faster in premenopausal women than in postmenopausal women, but the effect of menopause on clearance is not reversed by hormone replacement. Tirilazad clearance in these women is weakly related to midazolam clearance, a marker of CYP3A activity.

Phenytoin causes a rapid increase in 6 beta-hydroxycortisol urinary excretion in humans--a putative measure of CYP3A induction.

This study was conducted in eight healthy volunteers to assess the time course of induction of cytochrome P-450 by phenytoin. subjects received 200 mg of phenytoin every 8 h for 11 doses and 100 mg every 8 h for 8 doses. Trough concentrations of phenytoin in plasma were measured by HPLC. Urine samples were collected between 08:00 and 12:00 on days -1, 1, 2, 3, 4, 5, and 7. Urinary concentrations of 6 beta-hydroxycortisol (6 beta-OHC) and cortisol (C) were determined by HPLC and were reported as a ratio (6 beta-OHC/C); this ratio is a marker for the 3A isozyme family of cytochrome P-450 (CYP3A). Mean plasma phenytoin concentrations on day 7 were 15.4 +/- 7.20 micrograms/mL. Mean 6 beta-OHC/C ratios increased by a factor of 2.37 from baseline during the course of the study. Values for the ratios on days 4, 5, and 7 were significantly higher than baseline by Dunnett's test (one-sided) (p < 0.05); the day 3 value was borderline statistically significant. These results show that phenytoin rapidly induces the activity of the CYP3A family of isozymes, with effects apparent within 48 h after the initiation of phenytoin therapy.

Pharmacokinetics of tirilazad in healthy male subjects at doses above 6 mg/kg/day.

The dose proportionality of tirilazad pharmacokinetics at dosages above 6.0 mg/kg/day were assessed in 18 healthy male volunteers between the ages of 19 and 46 years. Subjects were randomized to receive either 1.5 mg/kg, 3.0 mg/kg, or 4.0 mg/kg tirilazad mesylate every 6 hours for 29 doses (daily doses of 6.0, 12.0, and 16.0 mg/kg/day for 7 days). Each drug dose was administered intravenously over 10 minutes. Plasma tirilazad, U-89678, and U-87999 (active reduced metabolites) were quantified by HPLC. Two subjects in the high dose group withdrew before the end of the study. Following the first dose of tirilazad, dose-corrected pharmacokinetic parameters for all 3 compounds did not differ significantly among dose groups. After the final tirilazad the mean half-life of tirilazad was approximately 80 hours. Mean apparent tirilazad clearance did not differ significantly among groups. Mean U-89678 AUC0-6 following the last tirilazad dose did not differ significantly between the 6.0 and 12.0 mg/kg/day doses, but the value for the 16.0 mg/kg dose was higher than values from both lower doses (p = 0.044 and 0.056, respectively). Similar results were obtained for U-87999. The dose effects observed for the pharmacokinetics of these 2 metabolites may have been a function of intersubject variability. When combined with previous data concerning the dose proportionally of tirilazad pharmacokinetics at doses less than 6.0 mg/kg/day, the data from the present study suggest that the pharmacokinetics of tirilazad are approximately linear over a dosage range of 1.0-16.0 mg/kg/day. Due to the inability to assess the plasma protein binding of tirilazad and its reduced metabolites, the clinical significance of the departure from linearity of the pharmacokinetics of U-89678 and U-87999 cannot be directly assessed. Further study at higher doses will be needed to address this issue.

Multiple-dose tolerability and pharmacokinetics of tirilazad mesylate at doses of up to 10 mg/kg/day administered over 5-10 days in healthy volunteers.

Multiple dose pharmacokinetics and tolerability of tirilazad mesylate were assessed at the maximum dosage and duration expected for tirilazad mesylate therapy of subarachnoid hemorrhage and head injury. Healthy male subjects (47) received either 1 mg/kg/day, 3 mg/kg/day, 6 mg/kg/day, or 10 mg/kg/day tirilazad mesylate, given as 10 minute i.v. infusions every 6 hours for 21 (at the highest dose) or 41 doses. Plasma tirilazad mesylate and U-89678, an active metabolite, were quantified by HPLC. Thirty-nine subjects completed the study. Tirilazad mesylate was generally well tolerated; injection site irritation was the primary adverse effect. Sporadic, dose unrelated elevations in liver enzymes were observed. All medical events were reversible. No clinically significant effects on vital signs, cardiac telemetry, or other laboratory values were seen. Both tirilazad and U-89678 accumulated on multiple dosing, and steady-state plasma levels were approximated by day 11 of dosing. U-89678 average steady-state concentrations approached 58% of those of the parent compound in the 6.0 mg/kg/day dose group. Following the last dose, mean half-lives for tirilazad ranged from 61.2-123 hours; mean U-89678 half-lives ranged from 60.5-111 hours. Tirilazad mesylate pharmacokinetics exhibited slight nonlinearity, AUC0-6 values for the 6 mg/kg/day were 33% higher than those predicted based on data from the 1.0 mg/kg dose group. Neither the long half-lives of U-89678 and tirilazad nor slight nonlinearity of tirilazad pharmacokinetics are likely to have significant clinical impact during short-term treatment of acute neurological injury.

Revamped, joint ventures make a comeback.

After a period of declining popularity, joint ventures are beginning to re-emerge as viable solution to challenges facing health care. Future joint ventures likely will be different, however, in response to problems and resulting legislation. Healthcare providers interested in starting a joint venture should consider five basic formats and some guidelines for developing them. Joint ventures ultimately may become a focal point for renewed efforts at regional healthcare planning.

Fixing the quick fixes to physician relations.

Traditional methods for cooperation among hospitals and physicians have not been successful at warding off problems that prompt hospitals and physicians to work together in the first place. Management service organizations and income guarantees have left hospitals open to third-party payer contracting problems, legal risks, and poor spending decisions. A solution perhaps begins with hospitals and physicians' groups finding a way to align their strategic planning processes.

Integrated delivery can ally physician and hospital plans.

Hospitals trying for successful business cooperation with physicians have become frustrated with risky joint ventures, quick-fix recruiting plans, and management service organizations. An integrated delivery system (IDS), however, may succeed where other approaches have failed. An IDS can align physician and hospital strategic planning, command more favorable contract terms from payers, and meet community needs in recruiting physicians and allocating capital. With the healthcare payment system likely to remain chaotic, however, the value of IDS improvements to healthcare delivery remains to be seen.

A guide to organizing joint ventures with physicians.

Catholic health care facilities must consider the business and legal risks, canon law, and other constraints when planning a joint venture with physicians. Participants should first establish goals and compatibility, then determine the venture's type (property, service), form ("true," lease, contract), and structure (corporation, partnership, joint property ownership, trust). The administrator must decide whether the facility will participate directly in the venture or form a separate organization. Participants must determine their relationships with the venture, choosing among many options. The administrator should consider whether a venture raises any canon law issues, especially regarding ecclesiastical and secular assets, approval by the local bishop or Holy See, and need for consultation. Other pertinent legal issues include: Fraud and abuse. The venture should not appear as compensation to induce referrals. Physician referrals. Many states prohibit or restrict referrals by physician participants. Antitrust law. Participants may be liable for actions constituting on antitrust violation. Securities low. Organizers must clarify Securities and Exchange Commission registration exemptions and observe state "blue sky" laws. Tax issues. Catholic health care facilities must consider such factors as tax-exempt status, unrelated business income, taxable subsidiaries, and public charity status. Other considerations include tax ramifications for physicians; tax shelter registration; certificate of need (CON), licensing, and building standards; effects on reimbursement and pension plans; organizational and bond documents; corporate medical practice and fee-splitting questions; and labor and contractual issues.

Population pharmacokinetics and pharmacodynamics of rivaroxaban in patients with non-valvular atrial fibrillation: results from ROCKET AF.

Two once-daily rivaroxaban dosing regimens were compared with warfarin for stroke prevention in patients with non-valvular atrial fibrillation in ROCKET AF: 20 mg for patients with normal/mildly impaired renal function and 15 mg for patients with moderate renal impairment. Rivaroxaban population pharmacokinetic (PK)/pharmacodynamic (PD) modeling data from ROCKET AF patients (n = 161) are reported and are used to confirm established rivaroxaban PK and PK/PD models and to re-estimate values of the models' parameters for the current AF population. An oral one-compartment model with first-order absorption adequately described rivaroxaban PK. Age, renal function, and lean body mass influenced the PK model. Prothrombin time and prothrombinase-induced clotting time exhibited a near-linear relationship with rivaroxaban plasma concentration; inhibitory effects were observed through to 24 hours post-dose. Rivaroxaban plasma concentration and factor Xa activity had an inhibitory maximum-effect (Emax ) relationship. Renal function (on prothrombin time; prothrombinase-induced clotting time) and age (on factor Xa activity) had moderate effects on PK/PD models. PK and PK/PD models were shown to be adequate for describing the current dataset. These findings confirm the modeling and empirical results that led to the selection of doses tested against warfarin in ROCKET AF.