Center for Drug Evaluation
and Research
Office of
Pharmacoepidemiology and Statistical Science
Office of Biostatistics
Statistical
Review and Evaluation
Clinical
Studies
|
NDA/Serial Number: |
21-686 |
|
Drug Name: |
Exanta (ximelagatran) |
|
Indication(s): |
• Prevention of VTE in patients undergoing knee
replacement surgery • Long term secondary prevention of VTE after standard
treatment for an episode of acute VTE |
|
Applicant: |
Astra Zeneca |
|
Date(s): |
Received
12/23/03 |
|
Review Priority: |
Standard
review |
|
Biometrics Division: |
Division
of Biometrics II |
|
Statistical Reviewer: |
Dionne L. Price, Ph.D. |
|
Concurring Reviewers: |
Stella
Grosser, Ph.D. S.
Edward Nevius, Ph.D. |
|
Medical Division: |
Division
of Gastrointestinal and Coagulant Drug Products |
|
Clinical Team: |
Ruyi
He, M.D. Kathy
Robie-Suh, M.D. |
|
Project Manager: |
Alice Kacuba |
|
|
|
|
|
|
|
Keywords: NDA review, clinical
studies |
|
|
|
|
|
|
|
|
|
|
Table
of Contents
1. EXECUTIVE SUMMARY
1.1 Conclusions and Recommendations
Astra
Zeneca has proposed Exanta (ximelagatran) tablets for the prevention of venous
thromboembolism (VTE)
in patients undergoing knee replacement surgery and for the long term secondary
prevention of VTE after standard treatment for an episode of acute VTE. The applicant claimed that ximelagatran
reduced the incidence of total VTE and/or mortality in patients undergoing total
knee replacement and that ximelagatran reduced the recurrence of VTE events
among patients who previously received 6 months of anticoagulation therapy for
an acute VTE. Based on my evaluation of
NDA 21-686, I concluded that statistical evidence supported the efficacy of
ximelagatran for the proposed indications.
However, several additional issues required consideration to completely
ascertain the strength of the efficacy results. The issues were beyond the scope
of this review and included the adequacy of the period of follow-up for the
short term indication, the appropriateness of warfarin as a comparator, and
concerns regarding the safety profile of the drug. These concerns are addressed in the clinical
review of Dr. Ruyi He.
1.2 Brief Overview of Clinical Studies
Support for the oral anticoagulant,
ximelagatran, was primarily derived from three studies. Two studies, EXULT A and EXULT B,
investigated the superiority of ximelagatran to warfarin in the prevention of
VTE. A single study, THRIVE III,
investigated the superiority of ximelagatran to placebo in the long term
secondary prevention of VTE.
EXULT
A and EXULT B were randomized, double-blind, multi-center studies conducted in
patients undergoing total knee replacement.
Eligible participants in the former study were randomized to
ximelagatran 24 mg, ximelagatran 36 mg, or warfarin. Patients in the latter study were randomized
to ximelagatran 36 mg or warfarin. Due
to varying appearances and dosing schedules, the treatments were administered
utilizing a double-dummy design. Patients received treatment twice daily for 7
to 12 days. The primary measure of
efficacy was the incidence of total VTE and/or all-cause mortality where total
VTE was defined as distal and/or proximal deep vein thrombosis and/or
symptomatic pulmonary embolism.
Treatment group differences were assessed using a
Cochran-Mantel-Haenszel chi-square test stratified by type of surgery.
THRIVE
III was a randomized, double-blind, placebo-controlled, multi-center study to
assess the efficacy of ximelagatran 24 mg given as a prolonged prophylaxis
after a six-month anticoagulation treatment.
The pool of eligible patients included individuals who had an
objectively verified, symptomatic VTE and had received anticoagulation treatment
for approximately six months. Two to
seven days prior to randomization, patients were instructed to terminate their
anticoagulation treatment. Study
participants were randomized to ximelagatran 24 mg or placebo administered
twice daily for 18 months. The primary
measure of efficacy was the time to the first VTE event. The analysis employed a log-rank test to
assess the treatment differences.
1.3 Statistical Issues and Findings
One
methodological issue arose during the course of my review. In
THRIVE III, analyses of the outcomes utilized the statistical methodology of
survival analysis. An underlying
assumption of the methodology, as applied by the applicant, is that all study
participants will experience a recurrence.
However in THRIVE III, a substantial number of participants may never
experience a recurrence; therefore, caution should be exercised when
interpreting conclusions from the planned analysis. I subsequently dichotomized the response and
performed an analysis similar to that of EXULT A and EXULT B. My analysis was post-hoc; however, the sole
purpose was to validate the conclusions.
The evidence
taken collectively from EXULT A and EXULT B indicated statistical support of
the efficacy of the ximelagatran for short-term use. When comparing ximelagatran 36 mg to warfarin,
a significant reduction in the incidence of VTE and/or all-cause mortality was
demonstrated in both studies. The
magnitude of the reduction was 7.3% and 9.3%
(ximelagatran 36 mg versus warfarin) in the two studies, respectively. A 24 mg dose of ximelagatran was additionally
included in EXULT A; however, the dose failed to demonstrate a significant
difference in the reduction of total VTE and/or all-cause mortality as compared
to warfarin. The evidence from THRIVE
III indicated statistical support of ximelagatran for the long term secondary
prevention of VTE. Specifically, ximelagatran 24 mg (administered twice daily for 18
months) significantly reduced the recurrence rate of VTE events compared to
placebo. The estimated cumulative risk
of a recurrent VTE was 2.8% for patients randomized to ximelagatran and 12.6%
for patients randomized to placebo.
2. INTRODUCTION
2.1 Overview
Astra Zeneca has proposed
Exanta (ximelagatran), an oral anticoagulant, for the following indications:
prevention of venous thromboembolism (VTE) in patients undergoing total knee
replacement, long term secondary prevention of VTE after standard treatment for
an episode of acute VTE, and prevention of stroke and other thromboembolic
complications associated with atrial fibrillation. According to the applicant, “The development
program for ximelagatran has been designed to offer an oral alternative
anticoagulant to VKAs for major indications.”
Exanta was introduced to the Division of Gastrointestinal and Coagulant
Drug Products through
IND 56, 611. The development plan was
discussed via several meetings and correspondences. Discussion topics focused on the
appropriateness of various clinically defined outcomes, the justification of
the proposed non-inferiority margin for specified studies, and the number and
type of studies required for the proposed indications. The current submission includes studies
investigating the superiority of ximelagatran to warfarin in prevention of VTE
in patients undergoing knee replacement, the superiority of ximelagatran to
placebo in the secondary prevention of VTE, and the non-inferiority of
ximelagatran to warfarin in the prevention of complications associated with
atrial fibrillation. This review will
focus on the former two indications only.
The latter indication will be reviewed by the Division of Cardio-Renal
Drugs.
2.2 Data Sources
Primary
support for ximelagatran for the prevention of VTE was derived from two
randomized, double-blind, multi-center studies, namely, EXULT A and EXULT
B. Support for the long term secondary
prevention indication was derived from THRIVE III, a randomized, double-blind,
multi-center study. The drug application
was electronic. The study reports and data were archived in the Food and Drug
Administration internal document room under the network path location
\\CDSEUB1\N21686\N_000\2003-12-23. A
summary of the studies is provided in Table 1.
Table 1: Table of studies
|
Study Number Number of centers
(n) |
Study Design |
Treatment Arms and
Number of randomized patients (n) |
Primary measure of
efficacy |
|
SH-TPO-0010 EXULT A Multi-center (114) |
Phase 3, randomized, double-blind, active-controlled study in patients undergoing total knee replacement |
•Exanta 24 mg (762) •Exanta 36 mg (775) •Warfarin (764) |
Composite endpoint of DVT and/or PE and/or all-cause mortality |
|
SH-TPO-0012 EXULT B Multi-center (113) |
Phase 3, randomized, double-blind, active-controlled study in patients undergoing total knee replacement |
•SKY0401 36 mg (1152) •Warfarin (1151) |
Composite endpoint of DVT and/or PE and/or all-cause mortality |
|
SH-TPO-0005 Multi-center (126) |
Phase 3, randomized, double-blind, active-controlled, , non-inferiority study in patients undergoing total hip arthroplasty |
•Exanta 24 mg (918) •Enoxaparin 30 mg (920) |
Incidence of overall VTE |
|
SH-TPO-0006 Multi-center (77) |
Phase 3, randomized, double-blind, active-controlled study in patients undergoing total knee replacement |
•Exanta 24 mg (348) •Warfarin (332) |
Incidence of overall VTE |
|
SH-TPO-0004 Multi-center (55) |
Phase 2, randomized, dose-finding, safety, and pharmacokinetics study in patients undergoing total knee replacement |
•Exanta 8 mg (85) •Exanta 12 mg (134) •Exanta 18 mg (126) •Exanta 24 mg (130) •Enoxaparin 30 mg (125) |
Incidence of overall VTE |
|
SH-TPV-0003 THRIVE III Multi-center (142) |
Phase 3, randomized, double-blind, placebo-controlled study in patients having received six-month anticoagulation treatment for VTE |
•Exanta 24 mg (617) •Placebo (616) |
Time to first VTE event |
3. STATISTICAL EVALUATION
3.1 Evaluation of Efficacy
The
main body of my evaluation of efficacy will discuss each study individually.
3.1.1
SH-TPO-0010(EXULT A)
Study Design and Endpoints
Due
to the varying appearances and dosing schedules, treatments were administered
utilizing a double-dummy design. After
surgery, eligible patients were randomized to one of the following three
treatment regimens:
- ximelagatran 24 mg
tablet with a 36 mg placebo tablet and placebo capsule(s) matching
warfarin,
- ximelagatran 36 mg with
a 24 mg placebo tablet and placebo capsule(s) matching warfarin, or
- warfarin 2.5 mg
capsule(s) and two placebo tablets matching ximelagatran 24 mg and 36
mg.
The
first dose of warfarin (and matching placebo) was administered on the evening
of the day of surgery, and subsequent doses were administered each
evening. Ximelagatran (and matching
placebo) was initially administered the morning after surgery. Subsequent doses
of ximelagatran were administered twice daily, in the morning and evening with
doses taken at 12 hour intervals. Of
note, the dose of warfarin was adjusted to maintain a target international
normalized ratio of 2.5 (range 1.8 to 3.0).
The duration of treatment varied from
The
primary measure of efficacy was incidence of total venous thromboembolism (VTE)
and/or all-cause mortality. Total VTE was defined as distal and/or proximal
deep vein thrombosis and/or symptomatic pulmonary embolism (with objective Independent
Central Adjudication Committee confirmation).
Deep vein thrombosis (DVT) was assessed using a venographic
procedure. The diagnosis of pulmonary
embolism (PE) was established by a ventral perfusion lung scan, pulmonary
angiography, or spiral computed tomography.
Additional details regarding the methodology used for assessments and
diagnoses of VTE are included in the appendix. Secondary measures of interest
included the incidence of proximal
Based
on previous studies, a sample size of 1700 was determined to be sufficient to
detect an approximate 7.5% difference in reduction in VTE between treatment
groups and warfarin with 80% power.
According to the applicant, the needed sample size was increased to 2250
to account for the percentage of patients (at most 25%) not having evaluable
venograms. Of note, 116 international
centers participated in the study; however, patients were only randomized at
114 centers.
Patient Disposition,
Demographic and Baseline Characteristics
Descriptive
demographic and baseline information was summarized using 2285 of the 2301
randomized patients. The excluded
patients did not receive study medication.
The ages of patients were between 32 and 89 with a mean age of 68. In the study, approximately 96% of the study
participants were Caucasian, and 4% of the participants were African American.
Females composed 62% of the patient population.
Baseline characteristics included weight, body mass index, nicotine use,
alcohol use, and creatinine clearance.
Demographic and baseline characteristics did not differ between
treatment arms. A detailed table
outlining the composition of the study population is presented in the appendix.
Of
the 2285 randomized patients, 1851 of the patients had surgery, received at
least one dose of the study medication, and were included in the efficacy
analyses. Approximately 98% of the
patients in the analysis population completed the study. Four patients discontinued due to adverse
events. Specifically, one patient in the
warfarin group, one patient in the ximelagatran 24 mg group, and two patients
in the ximelagatran 36 mg group discontinued due to an adverse event.
Statistical Methodologies
The
primary analysis employed a Cochran-Mantel-Haenszel chi-square test stratified
by type or surgery (i.e. bilateral or unilateral) to assess the treatment group
differences. Pairwise comparisons
between each dose of ximelagatran and warfarin were examined via a sequential
testing procedure to control the type I error rate. The ximelagatran 36 mg treatment arm was
initially compared to the warfarin arm.
The ximelagatran 24 mg group was subsequently compared to the warfarin
group if the previous comparison was significant at the 0.05 alpha level. The applicant additionally explored the
heterogeneity of the treatment effect across centers. According to the applicant, “Event rates and
non-evaluability rates were summarized by investigative site and the
possibility of treatment by center interaction was examined for sites with
large numbers of patients.” The analysis
of secondary endpoints was identical to the primary analysis; however, no
adjustments for multiple comparisons were made. A blinded interim analysis was
planned; however, the alpha inflation was considered negligible. Details of the
interim analysis are included in the appendix.
Analyses
were conducted on the intent-to-treat (ITT) population including all randomized
patients who had elective total knee replacement and who received at least one
dose of the study medication. The
applicant further stated, “No attempt was made to impute data for those
patients who did not have an evaluable venogram (as determined by central
adjudication) and did not experience a symptomatic, objectively confirmed VTE
or death. Thus, the efficacy ITT
analysis population included all patients who had (1) an evaluable venogram or
(2) symptomatic DVT/PE while being treated with study medication (objectively
confirmed by central adjudication) or (3) a fatality while being treated with
study medication.”
Results and Conclusions
Table
2 depicts results of the applicant’s primary analysis. Patients in the ximelagatran 36 mg group had
a significant reduction in the frequency of total VTE and/or all-cause
mortality as compared to patients in the warfarin group. The reduction in the frequency of the
composite endpoint in patients randomized to ximelagatran 36 mg was 7.3%
relative to patients randomized to warfarin.
Moreover, the relative risk reduction (i.e. risk difference expressed as
a percent of the risk in warfarin) was 26.4%.
A significant difference in the incidence of VTE and/or all-cause
mortality was not demonstrated between ximelagatran 24 mg and warfarin. For completeness, the applicant performed a
post-hoc analysis comparing the 24 mg and 36 mg doses of ximelagatran. There was no significant difference in the
composite outcome between the ximelagatran doses. These findings based on assessments by the
Independent Central Adjudication Committee (ICAC) were further supported by
assessments conducted locally at study centers.
Based on my independent evaluation of the data, I concur with the
primary results and conclusions.
Table 2: Frequency of total venous thromboembolism and/or all-cause
mortality-ITT population
(Source: Table 24, Clinical Study Report SH-TPO-0010)
I
additionally explored the individual events that composed the primary endpoint
to further elucidate the results in Table 2.
My results are illustrated in Table 3 and are in agreement with the
applicant’s results. Of note, asymptomatic DVTs were detected during mandatory
venography, and symptomatic DVTs were diagnosed by compression ultrasound and
confirmed using venograms. As evident
from the table, most events were asymptomatic, distal DVTs.
Table 3: Symptomatic and asymptomatic events over the treatment
duration -ITT population
|
Event |
Ximelagatran 24 mg |
Ximelagatran 36 mg |
Warfarin |
|
Number of subjects |
614 |
629 |
608 |
|
Asymptomatic DVT |
151 (25%) |
124 (20%) |
166 (27%) |
|
Proximal DVT |
12 (2%) |
13 (2%) |
23 (4%) |
|
Distal DVT |
139 (23%) |
111 (18%) |
143 (24%) |
|
Symptomatic DVT |
5 (.8%) |
7 (1%) |
9 (2%) |
|
Proximal DVT |
1 (.2%) |
2 (.3%) |
4 (.7%) |
|
Distal DVT |
4 (.7%) |
5 (.8%) |
5 (.8%) |
|
Pulmonary embolism |
2 (.3%) |
2 (.3%) |
0 (0%) |
|
Death |
1 (.2%) |
1 (.2%) |
1 (.2%) |
Per
the clinical team’s request, I also performed the primary analysis on the
safety population including all randomized patients who received at least one
dose of study drug. My results are shown
in Table 4.
Table 4: Frequency of
total venous thromboembolism and/or all-cause mortality-Safety population
|
|
Ximelag vs warfarin |
||||
|
Treatment group |
% (n/N) |
Exact 95% CI |
% |
95% CI |
CMH p-value* |
|
Ximelag 24 mg |
20.2 (153/757) |
(17.4, 23.1) |
-1.9 |
(-6.2, 2.3) |
0.692 |
|
Ximelag 36 mg |
16.6 (128/769) |
(14.0, 19.3) |
-5.5 |
(-9.6, -1.4) |
0.010 |
|
Warfarin |
22.1 (168/759) |
(19.2, 25.1) |
|
|
|
*
Treatment differences were tested using the Cochran-Mantel-Haenszel test,
adjusted for the type of surgery performed.
The
reduction in the frequency of the secondary composite outcome, incidence of
proximal
3.1.2 SH-TPO-0012(EXULT B)
Study Design and Endpoints
The
design and endpoints of study SH-TPO-0012 or EXULT B were nearly identical to
that of the previously described study (EXULT A) with variations in the tested
doses and sample sizes. In EXULT B, 2303
patients scheduled to undergo total knee replacement were randomized to
ximelagatran 36 mg or warfarin.
Based
on event rates in EXULT A, a sample size of 1720 was determined to be
sufficient to detect a 25% relative risk reduction in VTE with 90% power. According to the applicant, the needed sample
size was increased to 2300 to account for the percentage of patients (at most
25%) not having evaluable venograms. Of
note, 115 international centers participated in the study; however, patients
were only randomized at 113 centers.
Patient Disposition,
Demographic and Baseline Characteristics
The
ages of patients were between 26 and 91 with a mean age of 67. In the study, 94% of study participants were
Caucasian, and 5% were African American. Females composed 63% of the patient
population. Baseline characteristics
included weight, body mass index, nicotine use, alcohol use, and creatinine
clearance. Demographic and baseline
characteristics were similar across treatment groups. Detailed tables outlining the composition of
the study population with respect to demographic and baseline characteristics
are presented in the appendix.
Of
the 2303 randomized patients, 2299 received at least 1 dose of study
medication. Moreover of these 2299 patients,
1949 patients had a venogram adequate for evaluation and were included in the
analysis population. Approximately 99%
of the patients in the analysis population completed the study. Three patients in the warfarin group and six
patients in the ximelagatran group discontinued due to adverse events.
Statistical Methodologies
The
primary analysis assessed treatment group differences using a Cochran-Mantel
Haenszel test stratified by type or surgery (i.e. bilateral or
unilateral). Secondary endpoints were
analyzed using similar methodology.
Similar
to EXULT A, analyses were conducted on the intent-to-treat (ITT) population
consisting of all randomized patients who had elective total knee replacement
and who received at least one dose of the study medication. The applicant again stated, “No attempt was
made to impute data for those patients who did not have an evaluable venogram
(as determined by central adjudication) and did not experience a symptomatic,
objectively confirmed VTE or death.
Thus, the efficacy ITT analysis population included all patients who had
(1) an evaluable venogram or (2) symptomatic DVT/PE while being treated with
study medication (objectively confirmed by central adjudication) or (3) a
fatality while being treated with study medication.”
Results and Conclusions
The
results of the applicant’s analysis of the primary efficacy outcome as assessed
by the ICAC are depicted in Table 5.
Patients in the ximelagatran 36 mg group had a significant reduction in
the frequency of total VTE and/or all-cause mortality as compared to patients
in the warfarin group. The reduction in
the frequency of the composite endpoint in patients randomized to ximelagatran
36 mg was 9.3% relative to patients randomized to warfarin. Moreover, the relative risk reduction (i.e.
risk difference expressed as a percent of risk in warfarin) was 29.3%. Based on my independent evaluation of the
data, I concur with the results.
Table 5: Frequency of total venous thromboembolism and/or all-cause
mortality-ITT population
(Source: Table 15, Clinical Study Report SH-TPO-0012)
Table
6 further illustrates the individual components of the endpoint. Similar to
EXULT A, most events were asymptomatic, distal DVTs.
Table 6: Symptomatic and asymptomatic events over the treatment
duration -ITT population
|
Event |
Ximelagatran 36 mg |
Warfarin |
|
Number of subjects |
982 |
967 |
|
Asymptomatic DVT |
214 (22 %) |
301 (31 %) |
|
Proximal DVT |
30 (3 %) |
33 (3 %) |
|
Distal DVT |
184 (19 %) |
268 (28 %) |
|
Symptomatic DVT |
8 (1 %) |
15 (2 %) |
|
Proximal DVT |
2 (.2 %) |
1 (.1 %) |
|
Distal DVT |
6 (.6 %) |
14 (1.4 %) |
|
Pulmonary embolism |
2 (.2 %) |
5 (.5 %) |
|
Death |
4 (.4 %) |
2 (.2 %) |
Per
the clinical team’s request, I also performed the primary analysis on the
safety population including all randomized patients who received at least one
dose of study drug. My results are shown
in Table 7.
Table 7: Frequency of total venous thromboembolism and/or all-cause
mortality-Safety population
|
|
Ximelag vs warfarin |
||||
|
Treatment group |
% (n/N) |
Exact 95% CI |
% |
95% CI |
CMH p-value |
|
Ximelag 36 mg |
19.2 (221/1151) |
(17.0, 21.5) |
-7.6 |
(-11.1, -4.1) |
<0.001 |
|
Warfarin |
26.8 (308/1148) |
(24.3, 29.4) |
|
|
|
*
Treatment differences were tested using the Cochran-Mantel-Haenszel test,
adjusted for the type of surgery performed.
The
reduction in the frequency of the secondary composite outcome, incidence of
proximal
3.1.3 Additional
Studies
Three
additional controlled studies of the short-term use of ximelagatran were
conducted but were not of primary focus due to the varying patient populations
and/or doses. I will briefly describe
those studies for completeness of review of the NDA submission. SH-TPO-0005 was a randomized, double-blind,
active-controlled study of ximelagatran 24 mg and enoxaparin 30 mg for the
prevention of VTE following total hip arthroplasty. Treatments were administered twice daily for
7-12 days utilizing a double-dummy design.
Ximelagatran required oral administration while enoxaparin was
administered subcutaneously. The primary
measure of efficacy was the incidence of total VTE as confirmed by the
ICAC. According to the applicant,
“Statistical
non-inferiority was established if the upper confidence bound around the
between group difference in VTE frequency was less than 5%.” The study failed to demonstrate the
non-inferiority of ximelagatran to enoxaparin.
In addition to statistical concerns regarding the non-inferiority
margin, an indication in the hip arthroplasty population was not sought;
therefore, the study was not of focus during the current review.
The
applicant described SH-TPO-0004 as a “dose-finding, safety, and pharmacokinetic
study of
H 376/95 (ximelagatran) as a prophylaxis for thromboembolic complications after
total knee replacement.” Patients
undergoing knee replacement were randomized to 8, 12, 18, or 24 mg of
ximelagatran or enoxaparin 30 mg.
Ximelagatran was administered orally twice daily under blinded
conditions while enoxaparin was administered subcutaneously under open-label
conditions. The study did not
demonstrate a linear dose response among the four ximelagatran groups. Moreover, pairwise comparisons between each
of the ximelagatran groups and the enoxaparin group failed to show significant
differences in the incidence of VTE.
The
design of SH-TPO-0006 mimicked that of EXULT A and EXULT B with variations in
the treatment arms and the definition of the primary efficacy outcome. Patients were randomized to ximelagatran 24
mg or warfarin. The primary measure of
efficacy was the incidence of VTE. The
study failed to demonstrate a difference in the incidence of VTE between
ximelagatran and warfarin.
3.1.4 SH-TPV-0003(THRIVE III)
Study Design and Endpoints
Study
SH-TPV-0003 or THRIVE III was a double-blind, placebo-controlled, multi-center
study to assess the efficacy of ximelagatran 24 mg given as a prolonged
prophylaxis after a six-month anticoagulation treatment. The pool of eligible patients included
individuals who had an objectively verified, symptomatic VTE and had received
anticoagulation treatment for approximately six months. Two to seven days prior to randomization,
patients were instructed to terminate their anticoagulation treatment. Study participants were randomized to ximelagatran
24 mg or placebo administered twice daily for 18 months. “Study visits were scheduled at two weeks and
four weeks and then every month (three to five weeks) during the first six
months after randomization and thereafter every three months (10-14 weeks).”
Based
on assumed event rates of 6% and 2% in the placebo and ximelagatran arms
respectively, the applicant determined that a sample of size 1200 would be
sufficient to detect a difference with 90% power. One hundred and forty-two centers from eighteen
countries participated in the study.
Patient Disposition,
Demographic and Baseline Characteristics
The
ages of randomized patients ranged from 18 and 90 with a mean age of 57. In the study, 93% of study participants were
Caucasian, and females composed 47% of the patient population. Baseline characteristics included weight,
body mass index, nicotine use, alcohol use, and creatinine clearance. Demographic and baseline characteristics were
similar across treatment groups according to the applicant. Detailed tables outlining the composition of
the study population with respect to demographic and baseline characteristics
are presented in the appendix.
Of
the 1233 randomized patients, 10 patients did not have post-randomization data
and were excluded from the efficacy analysis.
Overall, 320 patients discontinued the study prematurely, and 129 of the
discontinuations were caused by adverse events.
Specifically, 144 patients within the ximelagatran group discontinued
prematurely, and 62 (43%) were caused by adverse events. One hundred and
seventy-six participants in the placebo group discontinued the study
prematurely. Of the 176 participants, 38% discontinued due to an adverse event
and 35% discontinued to the development of a study specific discontinuation
criteria.
Statistical Methodologies
The
primary measure of efficacy was the time to the first VTE event calculated as
the number of days from randomization until the occurrence of the first
relevant event. Secondary measures of
efficacy included the time until death by any cause, the time to locally
confirmed VTE events, and the time to the composite of VTE and/or death.
The
analyses employed log-rank tests to compare recurrence rates between treatment
and groups for the primary and secondary outcomes. Moreover, Kaplan-Meier estimates and hazard
ratios were employed to gain additional insight.
An
interim analysis was in the planned charter for the Safety Committee. The
analysis employed a Peto-type boundary for monitoring a positive trend in VTE
events. The alpha inflation was
considered negligible (" = 0.0476).
Results and Conclusions
Study
participants receiving ximelagatran 24 mg twice daily experienced a significant
reduction in recurrent VTE events as compared to patients receiving placebo. The estimated cumulative risk of a recurrent
VTE event at the end of 18 months was 2.8% for patients randomized to
ximelagatran and 12.6% for patients randomized to placebo. The hazard or risk of a recurrent VTE event
was smaller for participants in the ximelagatran group as evidenced by the
hazard ratio of 0.16. The results were
supported by the significant reduction in the secondary composite endpoint (VTE
and/or death) among the treated group as compared to the placebo group. In addition, the risk of death from all
causes was comparable among treatment arms.
The estimated cumulative risk over time of the primary endpoint is
displayed in Table 8. Figures 1 and 2
are graphical depictions of the cumulative risk versus time for the primary
endpoint and the secondary composite endpoint.
Table 8:
Estimated cumulative risk (%) of a VTE event, ITT population
(Source: Table 23, Clinical study report SH-TPV-0003)
Figure 1: Cumulative risk of
recurrent VTE events versus time, ITT population
(Source: Figure 5, Clinical Study Report SH-TPV-0003)
Figure 2: Cumulative risk of
recurrent VTE events and/or all-cause mortality versus time, ITT population
(Source: Figure 7, Clinical Study Report SH-TPV-0003)
Upon
exploration of the data, I found that 83 (71 in the placebo arm and 12 in the
ximelagatran arm) of the 1223 study participants experienced a recurrent VTE
event. Under this scenario, the analysis
employed by the applicant may result in misleading conclusions; therefore, I
performed a post-hoc categorical data analysis by dividing the population into
two more stringent categories. One
category included all patients having a VTE event and/or discontinuing the
study prematurely. All remaining
patients were classified as not having an event. My analysis demonstrated a significant
reduction in recurrence rates for participants treated with ximelagatran as
compared to participants randomized to placebo.
3.2 Evaluation of Safety
The
evaluation of safety is deferred to the review of Dr. Ruyi He.
4. FINDINGS IN SPECIAL/SUBGROUP POPULATIONS
For ease of exposition and
completeness, I have included all the subgroup factors considered by the
applicant in the following review of subpopulations. Although the applicant examined numerous prognostic
factors, I focused primarily on the age, gender, and race factors.
4.1 Gender, Race and Age
4.1.1
SH-TPO-0010(EXULT A) and SH-TPO-0012(EXULT B)
The
impact of various subgroup factors on the incidence of total VTE and/or
all-cause mortality was investigated by the applicant via frequency tables and
a logistic regression model. The
logistic regression model employed in EXULT A included the following factors:
treatment, gender, age, race, country, type of surgery, body mass index,
creatinine clearance, history VTE, type of anesthesia, time to first dose, and
time to ambulation. The logistic model
employed in EXULT B included an additional factor for time to venography. In
addition, the interaction of each subgroup factor with treatment was also investigated
in both studies.
In
EXULT A, the applicant concluded that females, older patients, and patients
enrolled in
I
additionally conducted separate analyses by gender, age, and race. I initially
explored the effect of each subgroup factor on the primary endpoint using logistic
regression models. In EXULT A, older patients (> 70) had slightly higher
odds of experiencing a VTE and/or death as compared to younger patients
(≤ 70). In EXULT B, older patients
(> 70) and female patients had slightly higher odds of experiencing a VTE
and/or death as compared to younger patients (≤ 70) and male patients,
respectively. I also investigated the
significance of the treatment effect after adjusting for covariates. The treatment effect was consistent across
the subgroups.
The
applicant did not propose any efficacy claims for any subgroup of patients.
Overall, the results were consistent and lend support to the findings presented
in the preceding sections.
4.1.2 SH-TPV-0003(THRIVE III)
The
impact of several prognostic factors on the time to the first VTE event was
investigated by the applicant via frequency tables and Cox regression
models. Potential prognostic
factors identified by the applicant
included sex, age, creatinine clearance, weight, initial VTE event, previous
VTE events, presence of malignancy, and prothrombotic state. Separate analyses for each prognostic factor
were conducted via Cox regression models.
The between treatment comparisons included treatment as a covariate in
the models while the within treatment comparisons included the subgroup factors
as covariates, respectively.
The
applicant’s analysis yielded a gender effect.
Specifically, females experienced a lower risk of recurrent VTE events
as compared to males. The treatment
effect remained constant across age, race, and gender. I additionally conducted analyses on the
dichotomized VTE outcome. The results of
my analyses are in agreement with those of the applicant.
The
applicant did not propose any efficacy claims for any subgroup of patients.
Overall, the results were consistent and lend support to the findings presented
in the preceding sections.
4.2 Other Special/Subgroup Populations
No
additional analyses on other subgroup populations were warranted.
5. SUMMARY
AND CONCLUSIONS
5.1 Statistical Issues and Collective Evidence
One
methodological issue arose during the course of my review. In THRIVE III, analyses of the outcomes
utilized the statistical methodology of survival analysis. This methodology uniquely handles “time to
event” data where observations may be censored due to discontinuation or to the
limitation of the study’s follow-up time.
An underlying assumption of the methodology, as applied by the
applicant, is that all study participants will experience a recurrence. However in THRIVE III, a reasonable
assumption is that not all study participants will experience a
recurrence. I subsequently
dichotomized the response and performed an analysis similar to that of EXULT A
and EXULT B. My analysis was post-hoc;
however, the sole purpose was to validate the conclusions.
The evidence taken collectively from EXULT A and
EXULT B indicated statistical support of the efficacy of the ximelagatran for
short-term use. When comparing
ximelagatran 36 mg to warfarin, a significant reduction in the incidence of VTE
and/or all-cause mortality was demonstrated in both studies. A 24 mg dose of ximelagatran was additionally
included in EXULT A; however, the dose failed to demonstrate a significant difference
in the reduction of total VTE and/or all-cause mortality as compared to
warfarin. The evidence from THRIVE III
indicated statistical support of ximelagatran for the long term secondary
prevention of VTE. Specifically, ximelagatran
24 mg significantly reduced the recurrence rate of VTE events compared to
placebo.
5.2 Conclusions and Recommendations
Astra Zeneca proposes ximelagatran for the prevention of VTE in patients undergoing knee replacement surgery and for the long term secondary prevention of VTE after standard treatment for an episode of acute VTE. The primary claims are that ximelagatran (36 mg) reduces the incidence of total VTE and/or mortality in patients undergoing total knee replacement and that
ximelagatran (24 mg) reduces the recurrence of VTE events among patients who previously received 6 months of anticoagulation therapy for an acute VTE. My review of the collective evidence suggests statistical support of the efficacy of ximelagatran; however, statistical significance does not imply clinical meaningfulness. Issues warranting additional consideration include the adequacy of the period of follow-up for the short term indication, the appropriateness of warfarin as a comparator, and concerns regarding bleeding and liver toxicity. Discussion of these issues as they pertain to the clinical meaningfulness of the reduction in VTE events and the safety profile of the drug are deferred to the clinical review of Dr. Ruyi He.
6. APPENDIX
6.1 EXULT A
Additional
details regarding the methodology used for assessments and diagnoses of VTE
follow. The details are relevant to the
statistical analyses as the variables were coded in the datasets according to
the classifications described. The
information is extracted from the clinical study report SH-TPO-0010 of NDA 21-686
(pages 47-49).
Demographic and baseline
characteristics
(Source: Table 16, Clinical Study report SH-TPO-0010)
Demographic and baseline
characteristics continued
(Source: Table 16, Clinical Study report SH-TPO-0010)
The
following information extracted from page 66 of the clinical study report
SH-TPO-0010 outlines the interim analysis.
Secondary
endpoint: Frequency of proximal
(Source: Table 25, Clinical Study Report SH-TPO-0010)
6.2 EXULT B
Demographic
and baseline Characteristics
(Source: Table 15, Clinical Study Report SH-TPO-0012)
Demographic
and baseline characteristics continued
(Source: Table 15,
Clinical Study Report SH-TPO-0012)
Secondary
endpoint: Frequency of proximal
(Source: Table 26, Clinical Study Report SH-TPO-0012)
6.3 THRIVE III
Demographic
and baseline characteristics
(Source: Table 12, Clinical Study Report SH-TPV-0003)
Demographic
and baseline characteristics continued
(Source: Table 12, Clinical Study Report SH-TPV-0003)
Demographic
and baseline characteristics continued
(Source: Table 12, Clinical Study Report SH-TPV-0003)
