CENTER FOR DRUG EVALUATION AND
`
`RESEARCH
`
`APPLICA TION NUMBER:
`
`22-387
`
`STATISTICAL REVIEW! S)
`
`

`

`Office ofBioslalislics
`
`US. Department of Health and Human Services
`Food and Drug Administration
`Center for Drug Evaluation and Research
`OlTice of Pharmacoepidemiology and Statistical Science
`
`STATISTICAL REVIEW AND EVALUATION
`
`CLINICAL STUDIES
`
`NDA/SerialNumber:
`
`22-387
`
`Drug Name:
`
`Indication(s):
`
`Applicant:
`
`Date(s):
`
`Inhaled treprostinil sodium
`
`Treatment of pulmonary arterial hypertension
`
`United Therapeutics Corporation
`
`June 27, 2008
`
`Review Priority:
`
`Standard
`
`Biometrics Division:
`
`Biometrics I (HFD-710)
`
`Statistical Reviewer:
`
`John Lawrence
`
`Concurring Reviewers:
`
`Jim Hung
`
`Medical Division:
`Clinical Team:
`
`Division of Cariorenal Drug Products (HFD-l 10)
`Abraham Karkowsky, Medical Reviewer
`
`Project Manager:
`
`Dan Bium
`
`Keywords: Nonparametric ANCOVA
`
`

`

`Table of Contents
`
`1.
`
`EXECUTIVE SUMMARY ...........................
`
`1.1
`I .2
`1.3
`
`CONCLUSIONS AND RECOMMENDATIONS ..........................................................
`BRIEF OVERVIEW OF CLINICAL STUDIES .........................................................................
`STATISTICAL ISSUES AND FINDINGS ............................................................................................................... 3
`
`
`
`2.
`
`INTRODUCTION ...............................................................................................................................................4
`
`OVERVIEW ......................................................................................................................................................4
`DATA SOURCES ...................................................................................................................
`
`
`
`2.]
`2.2
`
`3 .1
`3.2
`
`3.
`
`STATISTICAL EVALUATION ....................................................................................................
`
`EVALUATION OF EFFICACY ......................................................................................................................... 6
`EVALUATION OF SAFETY ............................................................................................................................... 9
`
`4.
`
`FINDINGS IN SPECIAL/SUBGROUP POPULATIONS ......................................................................... 10
`
`4.1
`4.2
`
`GENDER, RACE AND AGE .................................................................................................................... 10
`
`OTHER SPECIAL/SUBGROUP POPULATIONS .................................................................................................. I0
`
`5.
`
`SUMMARY AND CONCLUSIONS ................................................................................................................ 10
`
`5.1
`52
`
`STATISTICAL ISSUES AND COLLECTIVE EVIDENCE ....................................................................................... 10
`CONCLUSIONS AND RECOMMENDATIONS ..................................................................................................... I
`I
`
`

`

`1.
`
`EXECUTIVE SUMMARY
`
`1.1 Conclusions and Recommendations
`
`The primary endpoint of the study was change in exercise capacity at Week 12, as
`measured by the Peak 6MWD. A nonparametric analysis of covariance
`(ANCOVA), adjusted for baseline walk and disease etiology was performed for all
`patients.
`In the treated group, the median change from baseline was 22 m compared to a
`median improvement of3 m in the placebo group (p=0.004, estimated difference of 20 m
`with a 95% confidence interval of(8, 33). According to the sponsor's study report, the
`treatment was well—tolerated and has an acceptable safety profile.
`
`1.2 Brief Overview of Clinical Studies
`
`This clinical program included l phase 3 safety and efficacy study. This was an
`international, double blind, multi—center, randomized, placebo controlled, parallel—group
`study in NYHA Class III and IV adult patients with severe PAH on a stable dose of 125
`mg twice daily (BID) of bosentan or a stable dose of sildenafil for at least three months
`prior to study stait. Patients were required to have a Baseline six—minute walk distance
`(6MWD) of200-450 meters. This study consisted ofa Baseline Visit and a Treatment
`Period, for a total of 12 weeks. The primary endpoint was the placebo corrected change
`in distance walked at 12 weeks compared to Baseline in the peak six-minute walk
`distance (6MWD). Peak 6MWT was assessed no less than ten minutes and no more than
`60 minutes post study drug inhalation. All 6MWTs were scheduled to be performed 3-5
`hours after the bosentan dose or 30 to 120 minutes after the sildenafil dose. These time
`periods correspond to the peak plasma concentrations for each of these concomitant
`therapies.
`
`1.3 Statistical Issues and Findings
`
`The primary endpoint, median change in 6MWD was analyzed using a nonparametric
`analysis of covariance (ANCOVA), with adjustment for etiology and baseline 6M WD.
`In the treated group, the median change from baseline was 22 in compared to a median
`improvement of3 m in the placebo group (p=0.004, estimated difference of20 m with a
`95% confidence interval of(8, 33).
`
`The secondary endpoints were analyzed in the following pre—specified hierarchy to
`control the Type I error rate: time to clinical worsening, Borg dyspnea score, NYHA
`functional class, trough 6MWD at Week 12, Peak 6MWD at Week 6, quality of life,
`signs and symptoms of PAH, and Peak 6MWD at Day 1. Four subjects in the treated
`group and 6 subjects in the placebo group experienced clinical worsening. There was no
`difference in the time to clinical worsening between treatment groups (p=0.5829).
`
`Lu‘
`
`

`

`2.
`
`INTRODUCTION
`
`2.1 Overview
`
`This clinical program included 1 phase 3 safety and efficacy study. This was an
`international, double blind, multi—center, randomized, placebo controlled, parallel-group
`study in NYHA Class III and IV adult patients with severe PAH on a stable dose of 125
`mg twice daily (BID) of bosentan or a stable dose ofsildenafil for at least three months
`prior to study start. Patients were required to have a Baseline six-minute walk distance
`(6M WD) of200-450 meters. This study consisted ofa Baseline Visit and a Treatment
`Period, for a total of 12 weeks. The primary endpoint was the placebo corrected change
`in distance walked at 12 weeks compared to Baseline in the peak six—minute walk
`distance (6MWD). Peak 6MWT was assessed no less than ten minutes and no more than
`60 minutes post study drug inhalation. All 6MWTs were scheduled to be performed 3—5
`hours after the bosentan dose or 30 to l20 minutes after the sildenafil dose. These time
`periods correspond to the peak plasma concentrations for each of these concomitant
`therapies.
`
`The baseline demographics of the 235 patients studied are in Table 1. There do not
`appear to be any significant differences between the groups with respect to these
`variables. The mean age is about 55, the majority are female and NYHA Class 111.
`About 2/3 were taking bosentan and the remainder were taking sildenafil as background
`therapy.
`
`

`

`Table 1 Summary of demographics and other baseline data (ITT population).
`
`
`O n :4
`
`Inhaled TRF-
`Placebo
`‘
`
`Clnn'acterist‘ic
`
`
`(11:120)
`(11:115.)
`‘—
`52 {18—75)
`W (20-75)
`LAge in Yet-11's: 111ean(_1‘ange)
`Gender: Male .1 Female (11)
`22.."93
`33598
`
`l—FAH°- Etiology: n car-a)
`
`
`
`
`
`
`3.46 2‘ 63
`
`
`
`
`infiZSfi)
`5 J, (18- I 5
`/l 1.519] h‘l
`
`IPAH’
`CW):
`
`64 (56)
`40 (35)
`
`6? (56";
`:7 (311)
`
`13.1 (56)
`"7 (33:)
`
`Other
`1100)
`lGUSl
`27m)
`
`
`"7 (6?)
`
`38 (:5: )
`
`165 ( 70)
`
`70 ['30)
`
`Background PAH*
`
`Therapy: 11 ('36)
`
`Bosentan
`
`
`
`Sildenafil
`“t'ffiifie—on HEc—figEI—IIEF“
`
`Therapy:
`
`Mean Weeks : SD
`
`Bosentan
`
`Sildenafil
`
`
`
`Baseline :‘(YI-IA:Q Class:
`
`
`
`III .-" IV (11)
`
`Baseline GIVIVS-‘Dl:
`551132111: Q’D
`
`
`(meters)
`
`* E‘ulmoumg’ Amerial Hypenension
`
`- Idiopathic Pulmoumy Ax'zerinl Rwandan-Elm
`: Collagen '\"r1:»cularDiseas~e
`New Yeti: Hean Associmion
`"fi-Mimne Wall: Distance
`
`Source: p 53 of Study Report.
`
`

`

`For the primary efficacy analysis of change in 6MWD at Week 12 in the lTT population,
`a non-parametric analysis of covariance (ANCOVA) was performed, with adjustment for
`baseline walk and disease etiology. An ordinary least squares regression was fit to the
`observed changes from baseline in distance walked at Week 1 Peak, Week 6 Peak, Week
`12 Peak according to expanded windows and at Week 12 Peak according to the standard
`windows as a function of distance walked at baseline and disease etiology. Standardized
`ranks were calculated using the standardized residuals.
`lmputation was used for missing
`primary efficacy data. Lowest rank was assigned for death, discontinuation due to disease
`progression, and for patients in the study who initiated additional approved PAH therapy
`prior to the assessment. Last rank carried forward was assigned for adverse events or
`Withdrawal of consent, and other reasons for missing assessments. The standardized
`ranks were re—calculated after imputation. Finally, the Cochran-Mamel-Haenszel mean
`score statistic and p—value were calculated to compare these standardized ranks between
`treatment groups. The median difference between treatment groups was determined by
`the Hodges—Lehmann estimate.
`
`The secondary endpoints were analyzed in the following pre—specified hierarchy to
`control the Type I error rate: time to clinical worsening, Borg dyspnea score, NYHA
`functional class, trough 6MWD at Week 12, Peak 6MWD at Week 6, quality of life,
`. signs and symptoms of PAH, and Peak 6MWD at Day l. For clinical worsening, the p-
`value from the log-rank test was calculated. The changes in Borg dyspnea score would
`be compared using the Wilcoxon test. Changes in walking distance at other time points
`would be tested using similar methodology as the primary endpoint.
`
`2.2 Data Sources
`
`Electronic study reports (\\CDSESUB1\EVSPROD\NDA022387\0000) and data sets
`(\\cdsesub l \EVSPROD\NDA0223 87\0000\m5\datasets\lrx—triumph—OO 1 \analysis).
`
`3.
`
`STATISTICAL EVALUATION
`
`3.1 Evaluation of Efficacy
`
`For the primary efficacy analysis of change in 6MWD at Week 12 in the ITT population,
`a non-parametric analysis of covariance (ANCOVA) was performed, with adjustment for
`baseline walk and disease etiology. An ordinary least squares regression was fit to the ‘
`observed changes from baseline in distance walked at Week l Peak, Week 6 Peak, Week
`12 Peak according to expanded Windows and at Week 12 Peak according to the standard
`windows as a function of distance walked at baseline and disease etiology. Standardized
`ranks were calculated using the standardized residuals.
`Imputation was used for missing
`
`

`

`primary efficacy data. Lowest rank was assigned for death, discontinuation due to disease
`progression, and for patients in the study who initiated additional approved PAH therapy
`prior to the assessment. Last rank carried forward was assigned for adverse events or
`withdrawal of consent, and other reasons for missing assessments. The standardized
`ranks were re—calculated after imputation. Finally, the Cochran—Mantel-Haenszel mean
`score statistic and p~value were calculated to compare these standardized ranks between
`treatment groups. The median difference between treatment groups was determined by
`the Hodges-Lehmann estimate.
`
`In both groups, about 90% 0fthe subjects completed the study (see Table 2).
`
`Table 2 Patient disposition
`
`Treatment
`
`Study Disposition
`
`Tuhakthltccbn
`
`'
`
`V
`
`" Total
`
`
`3335
`11
`1'20
`n
`n 7; 115.
`
`Completed Study 11 (fits)
`
`102 (89)
`
`110 (92)
`
`212 (90)
`
`Treprostinil
`
`7 (6)
`
`
`Did Not Complete Study 11, (0/6)
`
`13 (11)
`
`Death
`
`\Vorsening PAH
`
`Adverse Event
`
`'
`
`,
`
`o
`
`3 (3:)
`
`Withdrawal of Consent
`
`3 (3.)
`
`Source: Study Report p. 48.
`
`In the treated group, the median change from baseline was 22 m compared to a median
`improvement of3 m in the placebo group (p=0.004, estimated difference of20 m with a
`95% confidence interval ol“(8, 33). Figure 1 shows a graph ofthe Observed Week 12
`Peak values on the y-axis vs. the baseline values on the x-axis. Figure 2 shows the
`empirical cumulative distribution function for the observed changes from baseline to
`Week 12 Peak in both groups. I am not sure if this should really be called empirical
`distribution functions because missing values were excluded, but they are the observed
`values on the x—axis and the proportion ofthe observed values less than or equal to x on
`the y-axis; there is ajump of size 1/11 at each observed value where n is the number of
`observed values in the group. Figure 3 shows boxplots ofthe standardized ranks of the
`residuals by treatment group, showing the active treatment group had a median larger
`than the placebo group.
`
`

`

`Figure 1 Scatterplot of baseline values and observed Week 12 Peak distance walked
`(FDA analysis).
`
`
`—,—to
`
`o0
`
`o
`
`—
`
`8
`m
`
`0‘
`\r
`o
`
`x
`‘1’
`m
`N o
`n.
`‘- 0-
`x (0
`(v
`a)
`g
`
`_
`
`o
`8
`
`o
`'—
`o—
`
`0-
`
`o
`
`o
`
`°
`§
`
`0
`
`00
`
`o
`
`O
`000
`o
`o
`o
`o
`6) o
`o
`g
`000
`oo o
`06706
`o 000
`00
`(9
`OO O
`
`0
`
`O
`
`o
`
`o
`
`0
`
`000
`O
`o
`
`o
`o
`
`O
`
`O
`
`o
`
`o
`
`8
`O O@
`8
`0 8 00
`0
`0
`
`o
`
`o
`o
`
`l
`200
`
`
`
`
`0°
`
`o
`
`0
`
`0°
`O
`
`00%
`0 o o
`O (9
`O
`0°
`meg
`0%
`o
`oo oo
`o
`0° 0
`oo ‘80
`O
`00
`o
`°8°§ ooogflnoo
`oo
`00
`‘91
`o
`o
`(m o
`0
`o
`0
`
`o
`
`o
`
`o
`
`l
`250
`
`“
`
`“_l
`l-"
`350
`300
`Baseline Distance Walked
`
`l
`400
`
`1
`
`l
`':
`
`I
`450
`
`Figure 2 Empirical cumulative distribution function for both groups change in 6MWD
`from baseline at Week 12 Peak (FDA analysis).
`
`
`|
`l l
`1.0 .._|_
`l
`
`
`
`
`
`Proportionlessthanorequaltox
`
`0.8
`
`6 L—
`0'.
`
`
`
`I
`
`
`
`-200
`
`O
`-10()
`Change from Baseline Distance Walked
`
`100
`
`

`

`Figure 3 Boxplots of standardized ranks of residuals from regression at Week 12 Peak
`(FDA analysis).
`
`
`
`
`
`0.
`
`
`
`%—2m E
`
`Active
`
`Placebo
`
`J
`
`There was no difference in the time to clinical worsening between treatment groups
`(p=0.5 829). Therefore, none of the remaining secondary endpoints were tested or could
`be called statistically significant. No measurable median treatment effect on Borg
`dyspnea score (Hodges-Lehmann estimate of 0.0) was detected at any of the assessment
`periods (p=0.998 and p=0.623 at Weeks 6 and 12). Overall, there were no statistically
`significant difference in change in NYHA class between treatment groups (p= 0.807 at
`Week 12). There were no changes in signs and symptoms of PAH at Week 12. There
`was a small change in 6 minute walking distance at Week 12 trough and at Week 6 peak,
`but no change on Day 2 at peak.
`
`3.2 Evaluation of Safety
`
`According to the sponsor's study report, the treatment was well-tolerated and has an
`acceptable safety profile. The most common adverse events in the treprostinil group
`were cough, headache, and nausea. Although less common, flushing and syncope also
`appeared to occur more often in the treprostinil group compared to placebo.
`
`

`

`4.
`
`FINDINGS IN SPECIAL/SUBGROUP POPULATIONS
`
`4.1 Gender, Race and Age
`
`The difference in change in walking distance appeared to be consistent across age and
`gender subgroups (see Table 3). No analysis by racial subgroup was provided, possibly
`because almost all subjects were Caucasian.
`
`1111111111(«Away5’me 1,1,1111111111111 1111111
`11111
`11111111111111M111 11111111111111.11111.111111111111111wq11 «111.1111 11;»
`1111111111111111111111~111W W’KNNWWQ’WNCWWwmwewsw~11wwwz .
`.
`Table 3 Patient disposition Sponsors analys1s ofprimary endpoint by subgroups.
`
`
`
`.
`
`WA
`AI AAAAA {AAAA 11:: 115», AAAAA: 11212131
`
`figJAABA5311 {AAAAA mm PAW A,”All
`U54?
`Aimflll’
`'
`1
`
`
`QAAAMAZA AA AAA,
`AA AAA 1
`1
`
`{JAMAA 3?? Amie: A. 3A, FAA-m A AA]
`
`5A1:
`gm”, {WWI A»: A3 HASAAA: AA AA]
`A5338 AAAAA:
`1111:1739! mfg-m 1115:1313}
`
`(Emmafiia mien
`Amt;- AmArAA AAA; AAA“: Assisi-:13: At: $33
`AAA; flfWflifl [mi-m: ("AAA AAAAALIA: AMA}
`
`Source: p 76 of Study Report.
`
`4.2 Other Special/Subgroup Populations
`
`The difference between treatment groups appeared to favor treprostinil in both North
`America and in the rest of the geographic regions combined (see Table 3).
`
`5.
`
`SUMMARY AND CONCLUSIONS
`
`5.1 Statistical Issues and Collective Evidence
`
`The primary endpoint, median change in 6MWD was analyzed using a nonparametric
`analysis of covariance (ANCOVA), with adjustment for etiology and baseline 6MWD.
`In the treated group, the median change from baseline was 22 m compared to a median
`improvement of 3 m in the placebo group (p=0.004, estimated difference of 20 m with a
`95% confidence interval of (8, 33).
`
`10
`
`

`

`The secondary endpoints were analyzed in the following pie—specified hierarchy to
`control the Type I error rate: time to clinical worsening, Borg dyspnea score, NYHA
`functional class, trough 6MWD at Week l2, Peak 6MWD at Week 6, quality oflife,
`signs and symptoms of PAH, and Peak 6MWD at Day 1. Four subjects in the treated
`group and 6 subjects in the placebo group experienced clinical worsening. There was no
`difference in the time to clinical worsening between treatment groups (p=0.5829).
`
`5.2 Conclusions and Recommendations
`
`The primary endpoint of the study was change in exercise capacity at Week 12, as
`measured by the Peak 6M WD. A nonparametric analysis of covariance
`(ANCOVA), adjusted for baseline walk and disease etiology was performed for all
`patients.
`In the treated group, the median change from baseline was 22 m compared to a
`median improvement of 3 m in the placebo group (p=0.004, estimated difference 01°20 m
`with a 95% confidence interval of (8, 33). According to the sponsor's study report, the
`treatment was well—tolerated and has an acceptable safety profile.
`
`11
`
`

`

`This is a representation of an electronic record that was signed electronically and
`this page is the manifestation of the electronic signature.
`
`John Lawrence
`4/7/2009 10:50:42 AM
`BIOMETRICS
`
`Jame s Hung
`4/7/2009 05:47:54 PM
`BIOMETRICS
`
`