`RESEARCH
`
`
`
`APPLICATION NUMBER:
`203496Orig1s000
`
`MEDICAL REVIEW(S)
`
`
`
`
`
`
`
`
`
`
`
`NDA#203496 SD#32
`
`Oral treprostinil
`Medical Review: Maryann Gordon, MD
`Date submitted: 6-17-2013
`
`Conclusions
`
`I agree that the combination studies with current drugs used for pulmonary hypertension show
`inconsistent results. Although the NDA reviewers f01md that file studies wifl1 oral treprostinil on
`top of background therapy (#301 and #308) did not demonstrate efficacy, there are numerous
`reasons for study failure including small sample size and/or poor administration ofbackgrOImd
`therapy.
`
`In conclusion, the sponsor of a new PAH drug is not required to show that their drug has
`additional eflicacy when used in combination with other agents approved for the same
`indication.
`
`Background
`
`Summary of Published Combination Therapy Trials with Vasodilator Therapies in PAH
`Eight randomized, double-blind, placebo-controlled combination therapy studies were conducted
`in WHO Group 1 PAH patients with 4 studies evaluating the addition of a prostacyclin to
`existing background therapy, two studies evaluating addition of a PDE-5 inhibitor, and two
`studies evaluating addition of an ERA. The studies were 12 to 26 weeks in duration with
`enrollment ofNYHA Functional Class 11 through IV patients (Table l).
`
`Tabb l
`
`. fihohctsTeshdiICoflilliln
`
`w Trids
`
`
`
`A summary of the 6-minute walk distance (6MWD) results fi'om these combination therapy t1ials
`is provided in Figure 1.
`
`Reference ID: 3371561
`
`
`
`leSmyqumMRmth-flnfi-MTHIB
`
`
`
`Study design:
`Study #301 and #308 randomized subjects in a 1:1 ratio to UT-15C: placebo.
`Study #302 randomized subjects in a 2:1 ratio to UT-lSC: placebo. Study #302 was a 12-
`week study. Studies #301 and #308 were 16-week study. Patients eligible for enrollment
`in the three studies is similar to previous study for PAH drugs. Patients must have
`evidence for PAH and no evidence of lefi sided disease. They must be able to perform a
`walk test with 6MWD to be between 100 and 450 meters
`
`Dose:
`
`All fllree studies were titration to either tolerance or to adequate efi'ect. Major difl'erences in the
`three studies were doses and titration algorithms and the available dose formulation for the study.
`In general, subjects were started on a low dose of UT-lSC with upward titration at 3 day
`intervals based on symptoms of PAH as well as tolerability. All doses were taken with food.
`
`The overall conclusions by the FDA reviewers regarding the eflicacy of the studies #301 and
`#308 are summarized by Dr. Norman Stockbridge in his review dated 10-23-2012.
`
`Neither study 301 nor 308 (with background) was statistically significant at the 16-
`week assessment. An unplanned analysis of their combined results showed differences
`of 4 m at 4 weeks, 5 m at 8 weeks (neither p<0.05), 10 m at 12 weeks, and 10 m at 16
`weeks (the latter two time points were nominally statistically significant at p<0.01).
`
`Reference ID: 3371561
`
`
`
`---------------------------------------------------------------------------------------------------------
`This is a representation of an electronic record that was signed
`electronically and this page is the manifestation of the electronic
`signature.
`---------------------------------------------------------------------------------------------------------
`/s/
`----------------------------------------------------
`
`MARYANN GORDON
`09/11/2013
`
`SHARI L TARGUM
`09/16/2013
`
`Reference ID: 3371561
`
`
`
`
`
`DIVISION OF CARDIO-RENAL DRUG PRODUCTS
`Divisional Memo
`
`
`NDA:
`203496 Treprostinil extended-release tablets for
`pulmonary arterial hypertension.
`Sponsor:
`United Therapeutics
`Review date: 22 March 2013
`
`
`Reviewer:
`N. Stockbridge, M.D., Ph.D., HFD-110
`Distribution: NDA 203496
`This memo conveys the Division’s recommendation to issue a second Complete
`Response letter for treprostinil extended-release tablets.
`This application has been the subject of review by Dr. Karkowsky (5 March 2013), CDTL
`for the original submission. There is a subsequent review (21 March 2013) by Dr.
`Papoian that concludes that further consideration of the carcinogenic potential of
`treprostinil diolamine/diethanolamine can be deferred until the ongoing carcinogenicity
`study is complete.
`The Division previously issued a Complete Response letter for this application (23
`October 2012). Below, I give the issues in that letter, the sponsor’s response, and my
`current thinking. There are no new studies of oral treprostinil.
`
`You were able to demonstrate an effect on 6-minute walk only in study 302. The
`effect in that study was quite small and of dubious clinical importance. The
`estimated mean effect probably exaggerates the true effect, as much of the effect
`seems to be attributable to how values are imputed to subjects missing week 12
`data. (This appears to have been less of an issue with inhaled treprostinil. In
`addition, we note our disagreement about how some subjects in study 302 were
`categorized for the purposes of imputation.)
`
`The sponsor does not dispute the overall treatment effect size, but points out that it is
`the same as with other formulations of treprostinil, faint praise indeed. This effect is
`achieved at peak (where plasma levels are 7 to 10 times levels at trough with twice daily
`dosing. At trough, a statistically significant effect was not demonstrated, but the
`nominal effect was 13 m, about half the effect at peak.
`In defense of the clinical significance of this effect, the sponsor says that it is similar to
`the effect of subcutaneous treprostinil on 6-minute walk, and subcutaneous treprostinil
`was able to avert clinical worsening in patients discontinuing Flolan. However, I am
`skeptical that oral treprostinil would recapitulate this benefit, as subcutaneous
`administration does not result in peak-trough excursions of 7- to 10-fold.
`The sponsor also points out that survival in open label use of oral treprostinil is similar
`to that of subcutaneous treprostinil and that of bosentan and better than that seen in
`historical data. While this is somewhat reassuring from a safety perspective, neither
`subcutaneous treprostinil nor bosentan have mortality claims based on these open-
`label, historically controlled data, and there is no basis for attributing such good
`outcomes to oral treprostinil either.
`
`D:\NDA\N203496 Treprostinil oral\TreprostinilDivMemo2.doc
`
`1
`
`Last saved
`11:13 Friday, March 22, 2013
`
`Reference ID: 3280989
`
`
`
`Divisional memo
`Treprostinil oral tablets
`
`
`
`NDA 203496
` (pulmonary arterial hypertension)
`
`The sponsor notes that in the long-term open-label study, only 19% of subjects add
`another vasodilator in the first year. I do not know how to interpret that observation,
`but I am skeptical that it reflects normalization of subjects’ symptoms on oral
`treprostinil.
`In the primary analysis of study 302, 21% of subjects on oral treprostinil and 14% of
`subjects on placebo had imputed values, with the differences being among subjects
`assigned average placebo rank (4% vs. 0%) and those assigned last rank carried forward
`(8% vs. 1%). The differences that resulted in net better rank on oral treprostinil
`probably reflect its poorer tolerability; clinical deterioration and death were similar on
`study drug and placebo.
`In addition, pre-specified sensitivity analyses that (a) carry forward last rank for all
`missing data, (b) analyze completers only, or (c) use data obtained post-withdrawal, all
`show similar effect sizes and nominal p-values. In addition, analyses based on the FDA
`reviewer’s opinion of cause for withdrawal or on Dr. Wittes’s “worst reasonable case” all
`retain a similar effect size and at least nominal statistical significance.
`Thus the sponsor shows that the results are not highly sensitive to the imputation
`process. Nevertheless, it is difficult to describe the ‘advantage’ in rank obtained because
`of poorer tolerability of oral treprostinil.
`
`You were unable to demonstrate an effect on time to clinical worsening in three
`phase 3 studies.
`
`While this comment in the first Complete Response letter was intended to note merely
`that no benefit existed of greater clinical importance than the effect on 6-minute walk,
`the sponsor again reminds us that subcutaneous treprostinil has the claim I noted
`previously. I again note skepticism that this can be expected to apply to oral
`treprostinil, and that skepticism can be expected to make its way into labeling were oral
`treprostinil to be approved.
`
`You were unable to show an effect on 6-minute walk in two well-powered studies
`(301 and 308) in which subjects were on background therapy with other, possibly
`more effective but certainly better tolerated vasodilators. Given the meager effect of
`treprostinil and its poor tolerability, it is difficult to name a clinical scenario in which
`use of oral treprostinil is appropriate.
`
`The sponsor does not refute the findings of studies 301 and 308, but they note that that
`40 and 45% of subjects in these studies were on both a PDE5 inhibitor and an
`endothelin receptor antagonist. They do not follow up with an analysis by background
`treatment.
`In response to characterization of oral treprostinil as poorly tolerated, the sponsor notes
`that 824 subjects have participated in open-label studies, of whom 641 remained on
`treatment at 1 year. Whether that constitutes good tolerability is a matter of
`perspective, but I concede that some people tolerate long-term use. I also agree that no
`novel toxicity was associated with the oral formulation, and that the oral formulation
`avoids formulation-specific problems with inhaled, intravenous, and subcutaneous
`administration.
`The sponsor responds to the challenge of naming a clinical scenario for use of oral
`treprostinil by again noting open-label, long-term use and the low uptake of additional
`
`D:\NDA\N203496 Treprostinil oral\TreprostinilDivMemo2.doc
`2
`
`
`
`Last saved
`11:13 Friday, March 22, 2013
`
`Reference ID: 3280989
`
`
`
`Divisional memo
`Treprostinil oral tablets
`
`
`
`NDA 203496
` (pulmonary arterial hypertension)
`
`therapy. I again note my skepticism that this is a reflection of benefit of oral treprostinil
`rather than a benefit of remaining in a study.
`The sponsor goes on to argue that oral treprostinil has been adequately shown to work
`in some definable setting, asserting that it should be approved for use in that setting
`(monotherapy). I disagree. When there were only a few such drugs, then it made sense
`to approve them without concern about their interactions (or the small effect). Now
`there are multiple drugs in multiple classes. The symptomatic effects of any of them are
`so small as to be indiscernible by individual patients against the background of the day-
`to-day variability in symptoms1; this is why it takes hundreds of subjects to detect a
`treatment effect. The magnitude of the effect matters here, and if a new product or new
`formulation cannot be shown to achieve a clinically important effect alone, it ought to
`be demonstrated to contribute to a meaningful effect; oral treprostinil has done neither.
`The sponsor might reasonably attempt another study with a regimen giving less
`fluctuation in exposure across the inter-dosing interval, but showing an effect only
`absent other background therapy is of dubious value.
`
`
`1 Had this formulation demonstrated effects on more important end points, like mortality or progression, one
`might feel differently about the clinical importance of a small effect.
`D:\NDA\N203496 Treprostinil oral\TreprostinilDivMemo2.doc
`3
`
`
`
`Last saved
`11:13 Friday, March 22, 2013
`
`Reference ID: 3280989
`
`
`
`---------------------------------------------------------------------------------------------------------
`This is a representation of an electronic record that was signed
`electronically and this page is the manifestation of the electronic
`signature.
`---------------------------------------------------------------------------------------------------------
`/s/
`----------------------------------------------------
`
`NORMAN L STOCKBRIDGE
`03/22/2013
`
`Reference ID: 3280989
`
`
`
` NDA 203496 Treprostinil diolamine Complete response 3/5/2013 page 1
`
`
`MEMORANDUM
`DEPARTMENT OF HEALTH & HUMAN SERVICES
`Public Health Service
`Food and Drug Administration
`Center for Drug Evaluation and Research
`________________________________________________
` February 26, 2013
`
`DATE:
`
` FROM: Abraham Karkowsky, M.D., Ph.D. Group Leader, Division of
`Cardiovascular and Renal Products, HFD-110.
`
`
` TO:
`
`
`
`
`
` Dr. Norman Stockbridge, M.D., Ph.D., Director, Division of
`Cardiovascular and Renal Products, HFD-110.
`
`
` SUBJECT Complete response recommendation for NDA 203496, Treprostinil
`diolamine (no TRADENAME approved).
`
`
`This memo is the Divisions response to United Therapeutics resubmission dated
`January 31, 2013for oral Treprostinil (UT-15C). The Division’s recommendation, that the
`oral formulation of Treprostinil diolamine not be approved for the indication
`
` as a BID dosing
`
`regimen, has not changed based on the above submitted rebuttal.
`
`Only one study (#302) in the oral treprostinil application appears to demonstrate a
`placebo-subtracted median effect on six-minute walk distance (6MWD) that differed
`from zero in a pulmonary artery hypertension (PAH) population. There were two failed
`studies in this population when oral treprostinil was used on-top of approved therapies for
`PAH. There was also a failed set of studies that demonstrated no benefit in walk distance
`when treprostinil, as monotherapy, was subcutaneously administered. In the positive oral
`study the effect on walk-distance is substantially dependent on how missing values were
`imputed. Including the results of the initial application where treprostinil was
`administered subcutaneously the one nominally positive study was clearly an outlier. The
`single nominally positive study # 302 is insufficient, given the other negative study to
`allow for approval of the oral formulation of treprostinil diolamine.
`
`
`In addition, the sponsor’s proposed dosing instructions is based on an ever-
`shifting change in study protocol dosing strategy with lower doses and smaller dose
`increments recommended by the sponsor as the population demonstrated intolerance to
`the ongoing dosing regimens. UT-15C appears to be poorly tolerated and it is unlikely to
`be used as initial therapy. The two on-top of studies were unsuccessful and it would be
`difficult to recommend that UT-15C be used as adjunctive therapy to other PAH drugs.
`The pooled studies P01:04 and P01:05 utilizing subcutaneous infused treprostinil were as
`monotherapy. It is, therefore, not convincing to ascribe the failed oral studies were as a
`consequence of the use on top of concomitant therapies.
`
`
`Reference ID: 3271459
`
`(b) (4)
`
`
`
` NDA 203496 Treprostinil diolamine Complete response 3/5/2013 page 2
`
`
`The current dosing regimen utilized in the clinical studies was a BID regimen.
`That decision seems to not be based on the performance characteristics of the oral
`formulation. Based on the degree of excursion from peak concentrations from Cmax to C12
`hours, a BID regimen does not afford adequate concentrations during the entire dosing
`interval. The FDA, based on pharmacokinetic considerations, would recommend a TID
`dosing regimen for the UT-15C. However, no available information on 6WMD is
`available utilizing a TID regimen. Any new study should, in addition, better define a
`dose-response effect as well as the starting dose and appropriate dose increments.
`
`The two aspects, the large number of discontinuations requiring imputation of
`6MWD values and the lack of an appropriate dosing set of instructions are likely to be
`intimately related. Should an appropriate dose regimen been used in the clinical trials, the
`number of discontinuations and particularly the number of early discontinuations would
`have been reduced and the impact of imputation of missing data would have been
`minimized.
`
`Brief development summary of oral treprostinil:
`
`
`There were three placebo-controlled studies submitted to support a benefit of
`treprostinil diolamine (UT-15C) oral formulation. Two of these studies (#301 and # 308)
`were on top of accepted therapies for PAH, either phosphodiesterase 5 (PDE-5)
`inhibitors, endothelin receptor antagonists (ERA) or both. These two studies were on face
`value unsuccessful in demonstrating a benefit in 6MWD. Pooling of the two unsuccessful
`oral formulation studies was never pre-planned and moreover, because of the
`disproportionate numbers of dropouts in the treated group, the p-values for each of these
`studies are likely to be inaccurate and favor treatment group.
`
` The one nominally successful study (study #302) was a placebo-controlled study
`of UT-15C versus placebo in a pulmonary arterial hypertension (PAH) in the absence of
`approved background therapies. So, it becomes difficult to recommend the use of this
`drug with concurrent approved therapies.
`
`
`There were a large number of discontinuations during the one successful study,
`with many discontinuations occurring early in therapy and prior to any assessment of the
`effect of drug on 6MWD. The outcome is, therefore, largely dependent on the imputed
`values for these subjects. It does not appear that the subjects were followed for events of
`interest after the subject discontinued.
`
`
`Based on the discussion contained in the joint clinical/statistical review and the
`amendment dated 10-10-12, any benefit the use of UT-15C to increase walk distance in
`patients with pulmonary arterial hypertension (PAH) population is small, and toxicities
`are not trivial. There were a disproportionate number of dropouts in the UT-15C group
`early on in treatment in the one study which suggested a benefit (study #302). The
`assessment of the drug-effect markedly depends on the handling of these dropouts. The
`imputation rules are shown in Table 1. Several sensitivity analyses were contained in the
`
`Reference ID: 3271459
`
`
`
` NDA 203496 Treprostinil diolamine Complete response 3/5/2013 page 3
`
`
`joint clinical/statistical review and amendment suggest that the walk distance is very
`sensitive to the nature of the imputation of the missing values.
`
`Table 1: Missing data imputation algorithm study # 302
` Lowest rank
`For deaths, discontinuations due to clinical deterioration,
`transplantation or atrial septostomy or subject to ill to
`perform walk test
`For subjects who withdrew prior to any 6MWD
`For subjects who prematurely withdrew.
`
`Mean rank
`LOCF
`
`With respect to the oral formulation of treprostinil (treprostinil diolamine), there
`
`were three studies #301, study # 302 and study # 308. The results of the sponsor’s
`primary analysis are shown below. Also included are the results of the pooled
`subcutaneous therapy.
`
`Study design:
`Study #301 and #308 randomized subjects in a 1:1 ratio to UT-15C: placebo.
`Study #302 randomized subjects in a 2:1 ratio to UT-15C: placebo. Study #302 was a 12-
`week study. Studies #301 and #308 were 16-week study. Patients eligible for enrollment
`in the three studies is similar to previous study for PAH drugs. Patients must have
`evidence for PAH and no evidence of left sided disease. They must be able to perform a
`walk test with 6MWD to be between 100 and 450 meters
`
`Dose:
`
`All three studies were titration to either tolerance or to adequate effect. Major
`differences in the three studies were doses and titration algorithms and the available dose
`formulation for the study. In general, subjects were started on a low dose of UT-15C with
`upward titration at 3 day intervals based on symptoms of PAH as well as tolerability. All
`doses were taken with food. The table below defines the evolution of the dosing regimens
`used in each of the studies.
`
`Table 2: Dosing changes during placebo-controlled oral studies.
`Study
`Initial formulations
`Amendments altering dosing
`#
`available
`(all doses were administered BID)
`#301
`1, 5 mg
`0.5 mg and 0.25 mg amendments
`3 and 4, respectively
`
`0.5 mg and 0.25 mg amendments
`3 and 4, respectively.
`Amendment 6 introduced 0.125
`mg dose
`
`Dosing increments at
`final
`Dose was escalated
`every 3 days.
`
`Dose escalation was 1
`mg every 5 days but
`changed to 0.5 mg
`every 3 days.
`Amended to 0.25 mg
`every 3 days
`
`
`#302
`
`1, 5 mg
`
`Dosing increments initial
`
`Initial dose was 1 mg.
`Amendment 4 lowered
`starting dose to 0.5 mg
`Initial dose was 1 mg.
`Amendment 4 lowered
`starting dose to 0.5 mg.
`further lowered to 0.25 mg
`
`#308
`
`0.25, 0.5 and 1 mg. Aside
`from sites in China a 0.125
`mg dose became available.
`
`Initial dose was 0.25 mg. Initial
`increments was 0.25 mg prior to
`week 4. After week 4 the dose
`increments could be either 0.25 or
`0.5 mg increments. A 0.125 mg
`dose was allowed if the 0.25 mg
`dose increase was not tolerated
`
`Initial dose was 0.25 mg.
`Initial increments was
`0.25 mg prior to week 4.
`After week 4 the dose
`increments could be either
`0.25 or 0.5 mg increments.
`A 0.125 mg dose was
`allowed if the 0.25 mg
`dose increase was not
`tolerated
`
`Reference ID: 3271459
`
`
`
` NDA 203496 Treprostinil diolamine Complete response 3/5/2013 page 4
`
`
`
`
`The demographics of those enrolled are shown in Table 12.
`
`
`Table 3: Demographics of study #301, #302 and #308
`Parameter
`Study #301
`
`UT-15C
`N
`174
`Age
`51.1
`Gender (% female)
`85%
`Race: %
`
`placebo
`176
`49.5
`80%
`
`
`68%
`24%
`6%
`2%
`
`
`65%
`28%
`3%
`2%
`
`
`113 (65%)
`49 (28%)
`11 (6%)
`1 (<1%)
`
`
`
`
`119 (68%)
`43 (24%)
`11 (6%)
`3 (2%)
`
`placebo
`116
`43
`78%
`
`
`Study # 302
`UT-15C
`233
`41
`74%
`
`
`41%
`4%
`47%
`0
`
`41%
`<1%
`48%
`0
`
`171 (72%)
`48 (19%)
`12 (6%)
`2 (1%)
`< 1%
`
`
`
`None
`
`88 (76%)
`22 (19%)
`5 (4%)
`1 (<1%)
`0
`
`
`
`
`
`None
`
`
`
`
`
`Study # 308
`UT-15C
`157
`52
`76%
`
`
`placebo
`153
`50
`80%
`
`
`67%
`7%
`26%
`2%
`<1%)
`
`104 (66%)
`48 (31%)
`3 (2%)
`2 (1%)
`
`25 (16%)
`67 (43%)
`65 (41%)
`
`
`
`
`
`
`
`
`
`
`
`
`
`63%
`65
`29%
`3%
`0
`
`99 (65%)
`49 (32%)
`1 (<1%)
`4 (3%)
`
`28 (18%)
`65 (42%)
`60 (39%)
`
`
`90 (39%)
`143 (61%)
`
`
`43 (37%)
`73 (63%)
`
`
`43 (27%)
`113 (72%)
`
`
`37 (24%)
`115 (75%)
`
`Caucasian
`African/American
`Asian
`Native American
`Other
`
`Etiology
`Idiopathic/familial
`CVD
`Repaired CHD
`HIV
`Other
`Background therapy
`ERA
`PDE5-I
`Both
`Functional class
`II/III
`Functional class
`I-II
`III-IV
`
`
`
`
`55 (32%)
`45 (26%)
`74 (43%)
`
`41 (24%)/127 (76%)
`
`
`
`
`51 (29%)
`43 (24%)
`82 (47%)
`
`31/139
`
`
`The most notable differences between the three studies were the large fraction of
`Asians enrolled into study # 302. There was a 2:1 randomization in study #302. Studies
`#301 and #308 include usage of concomitant therapies. More than half the patients had
`functional class III or worse.
`
`Disposition of subjects:
`
`
`The disposition of patients in each of the studies is shown in the table below.
`A greater fraction among those entering study # 302 discontinued prematurely had no
`measurements on therapy and a substantial fraction of the patients had values imputed.
`
` Table 4 Completion status studies # 301, # 302 and # 308
`
`Study #301
`Study #302
`
`UT-15C
`UT-15C
`N=
`174
`233
`Completed study
`153 (88%)
`182 (78%)
`21 (12%)
`51 (22%)
`D/C prematurely
`Consent Withdrawn
`13 (7%)
`3 (1%)
`Death
`3 (2%)
`10 (4%)
`Lost to follow up
`2 (1%)
`4 (2%)
`Other
`2 (1%)
`4 (2%)
`Protocol Violation
`1 (<1%)
`
`Adverse event
`23 (10%)
`Clinical deterioration
`7 (3%)
`
`Placebo
`176
`167 (95%)
`9 (5%)
`3 (2%)
`2 (1%)
`1 (<1%)
`3 (2%)
`0
`
`Placebo
`116
`98 (84%)
`18 (16%)
`
`
`6 (5%)
`
`2 (2%)
`
`3 (3%)
`7 (6%)
`
`Study # 308
`UT-15C
`157
`132 (84%)
`25 (16%)
`1 (<1%)
`2 (1%)
`0
`
`
`18 (11%)
`4 (3%)
`
`Placebo
`153
`138 (90%)
`15 (10%)
`2 (1%)
`3 (2%)
`1 (<1%)
`
`
`5 (3%)
`4 (3%)
`
`
`
`Reference ID: 3271459
`
`
`
` NDA 203496 Treprostinil diolamine Complete response 3/5/2013 page 5
`
`
`The completer status and nominal reason for discontinuation for the three
`
`controlled studies is shown in Table 4.
`
`
`In each of the studies there was a greater fraction of patients who were allocated
`to UT-15C who discontinued for various reasons. Between 6-7% more patients treated
`with UT-15C discontinued than those who were treated with placebo.
`
`
`Statistical plan:
`The oral treprostinil studies were analyzed by an analysis of covariance, adjusted
`for baseline walk distance and when appropriate PAH background therapy. The
`magnitude of the treatment effects was defined by the Hodges-Lehmann method to
`estimate the median difference between treatment groups for the change from baseline in
`6MWD. Missing values were imputed by the algorithm in Table 1.
`
`
`The results of the three placebo-controlled studies for the pivotal walk distance
`metric are shown below. Study # 301 and # 308 were not statistically significant. The
`number of discontinuations (see Table 6) for the UT-15C subjects the numbers with
`imputed values was much greater than those in the placebo group.
`
`Results:
`
`
`P-01:04-P01:05
`
`
`
`
`10 NS (from label)
`
`
`The results of the 6 MWD data from the three UT-15C oral studies and from the
`pooled subcutaneous studies are shown below. Study # 302 appears as the clear outlier,
`
`Table 5: 6MWD placebo-controlled studies (meters)
`
`Study 301
`Study # 302
`PDE-5, ERA
`4 (-2, 12)
`9 (0, 18)
`
`13 (3,23)
`11 (0,22) NS
`
`Other therapy?
`Week 4
`Week 8
`Week 11 trough
`Week 12
`Week 16
`
`Consequence of Imputation:
`Please refer to the addendum Joint Clinical/Statistical review dated October 13,
`2012. There were two aspects that were addressed by this amendment. The first is the
`large difference in the number of subjects who required imputed values as a consequence
`of dropouts during the study. The second difference is the difference in rank between the
`UT-15C group and placebo among those where a value was imputed.
`
`
`No
`
`14 (4, 25)
`20 (7, 34)
`17 (3, 33)
`25.5 (10, 41) p< 0.001
`
`
`Study # 308
`PDE-5, ERA
`3(-4, 10)
`1 (-9, 11)
`
`6 (-5, 19)
`10 (-2 22) NS
`
`No
`
`Each subject who enrolled in the study is assigned a rank at the end of the study
`based on the walk-distance or the imputed values. The ranks ranged from best (1.0 to
`worst 0.0). There were 59 subjects who had no 12-week walk test in the UT-15C group
`versus 18 in the placebo group (note there was a 2:1 randomization scheme UT-15C:
`placebo). Imputing worst outcome values either for the UT-15C population alone or all
`patients regardless of therapy totally p=0.92 to p=0.21, respectively).
`
`
`Reference ID: 3271459
`
`
`
` NDA 203496 Treprostinil diolamine Complete response 3/5/2013 page 6
`
`
`In considering the12 -week data, the imputed rank for the 59 subjects who were in
`the UT-15C group and had no 12-week data was 0.36. The corresponding imputed rank
`score for the 19 placebo subjects was 0.11. So there appears to be a benefit simply
`because the imputed values in the UT-15C group were better than those of the placebo
`group.
`
`
`The statistician attempted another approach. He utilized a multiple imputation
`method to assign the missing data. For each subject who died the worst rank was imputed
`in this analysis (as for other analyses). The statistician assumed that those who did not
`walk would generally fall in the lower quartile if forced to exercise. For subjects with
`missing 6MWD test, a random value between 0 to 0.25 was assigned as well as some
`variance. The mean imputed value that was assigned was 0.125. The uncertainty in the
`assessment of rank is captured by the imputed variance. The imputed values with the
`sponsor’s analysis had a value of 0.45 for the UT-15C group and 0.16 for the placebo
`group. The result of the multiple imputation analysis yields p-value slightly > 0.05.
`
`As concluded by the Joint Medical/statistical review:
`“In summary, the robustness of the efficacy results depends heavily on
`how the missing data are treated in the statistical analysis; the p-value
`range from 0.0001 (from the sponsor’s analysis) to 0.92 (from analysis
`giving all treatment subjects with missing data the worst score). So, in
`my opinion the efficacy of treprostinil tablets has not been convincingly
`demonstrated, based on this study.”
`
`
`Safety:
`
`Some selective adverse events from the three studies are shown below. The three
`general classes of adverse events that appear to be more frequent in the UT-15C group
`than the placebo group reflect drug action and are vasodilatation, gastrointestinal
`symptoms and bone-muscle-joint pain. Some of these events are so much more common
`in the UT-15C group that blinding of the treatment may not be protected.
`
`Table 6: Adverse events in the categories of vasodilatation, gastrointestinal and prostacyclin-related.
`
`Study # 301
`Study # 302
`Study # 308
`UT-15C
`Placebo
`UT-15C
`Placebo
`UT-15C
`Placebo
`
`
`
`
`
`
`
`Vasodilatation:
`Headache
`Flushing
`Dizziness
`Gastrointestinal
`Nausea
`Diarrhea
`Vomiting
`Abdominal distention
`Decreased appetite
`Abdominal pain
`Prostacyclin-related
`Pain in jaw
`Pain in extremity
`Myalgia
`Arthralgia
`Back pain
`
`
`
`
`
`150 (86%)
`85 (49%)
`30 (17%)
`
`112 (64%)
`106 (61%)
`76 (43%)
`11 (6%)
`
`74 (43%)
`54 (31%)
`24 (14%)
`18 (10%)
`13 (7%)
`
`
`
`
`
`65 (37%)
`27 (15%)
`28 (16%)
`
`60 (34%)
`48 (27%)
`14 (8%)
`11 (6%)
`
`21 (12%)
`17 (10%)
`6 (3%)
`4 (2%)
`10 (6%)
`
`
`
`
`
`
`160 (69%)
`50 (21%)
`
`
`36 (31%)
`9 (8%)
`
`91 (39%)
`86 (37%)
`57 (24%)
`11 (5%)
`19 (8%)
`31 (13%)
`
`59 (25%)
`44 (19%)
`24 (10%)
`15 (6%)
`14 (6%)
`
`
`25 (22%)
`21 (18%)
`19 (16%)
`4 (3%)
`5 (4%)
`9 (8%)
`
`8 (7%)
`9 (8%)
`5 (4%)
`4 (3%)
`4 (3%)
`
`112 (71%)
`55 (35%)
`30 (19%)
`
`73 (46%)
`87 (55%)
`33 (21%)
`
`
`61 (40%)
`16 (10%)
`15 (10%)
`
`
`34 (22%)
`38 (25%)
`16 (10%)
`
`
`
`
`39 (25%)
`27 (17%)
`18 (11%)
`12 (8%)
`12 (8%)
`
`10 (7%)
`11 (7%)
`10 (7%)
`9 (6%)
`6 (4%)
`
`Reference ID: 3271459
`
`
`
` NDA 203496 Treprostinil diolamine Complete response 3/5/2013 page 7
`
`
`
`
`Other treprostinil therapies:
`There are currently two approved dosing forms of treprostinil. Treprostinil
`(Remodulin®) is available for parentral use administered either as a subcutaneous or
`intravenous infusion. The intravenous route was recommended for those who do not
`tolerate the drug by the subcutaneous route usually because of subcutaneous pain. The
`intravenous route was approved based on its kinetic equivalence to the use of
`Remodulin® by the subcutaneous route. Treprostinil (Tyvaso®) is also approved as an
`inhaled formulation. The steady state dose for the inhaled formulation was 9 breaths
`administered QID.
`
`
`Treprostinil when administered as a parentral formulation did not meaningfully
`alter walk distance in the pivotal studies that were performed for the initial NDA
`submission (2001). Moreover the point estimate for walk distance was small and again
`largely dependent on the handling of early dropouts and discontinuations.
`
`The sponsor specified an algorithm for imputing values in P01:04-P01:05 pooled
`studies. This algorithm is shown below (Table 2). This algorithm allowed imputation of
`values which were dependent on the investigator’s assessment of the reason for
`discontinuation. There are several hazards for accepting this imputation strategy. Since
`there were many more subjects in the treatment group than in the placebo group, the
`imputation algorithm alters the assessment of that population more than the placebo
`population.
`
` A
`
` higher dropout rate that required imputation of values, biases the result in favor
`of that group with a higher discontinuation /dropout rate. A person who discontinues
`early for adverse event or withdraws consent can never have a worst outcome attributed
`to the patient. Second, the process of discontinuing from a study is an integrative
`assessment of how the subject feels. If the subject is doing poorly on drug even a modest
`adverse event would be sufficient to discontinue study. So, hidden under the rationale for
`the reas