`
`CENTER FOR DRUG EVALUATION AND
`RESEARCH
`
`
`APPLICATION NUMBER:
`203496Orig1s000
`
`STATISTICAL REVIEW(S)
`
`
`
`
`
`
`
`
`
`Treprostinil tablets – NDA 203,496 Page 1 of 3
`⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯
`
`JOINT CLINICAL/STATISTICAL REVIEW
`(ADDENDUM)
`
`
`
`
`
`
`
`
`
`203,496
`United Therapeutics
`Treprostrinil tablets
`Pulmonary arterial hypertension
`October 23, 2011
`John Lawrence, Ph.D. (HFD–710)
`Maryann Gordon, M.D. (HFD–110)
`
`
`NDA #:
`
`
`Applicant:
`
`
`Name of Drug:
`
`Indication:
`
`
`Date of submission:
`Statistical Reviewer:
`Medical Reviewer:
`
`
`The purpose of this addendum is to provide some additional analyses of the pivotal Phase
`3 efficacy study TDE-PH-302. As noted on p. 24 of the original review, at the end of
`study (Week 12), the estimated placebo-subtracted change in 6-minute walking distance
`was 25.5 m (p=0.0001) using the sponsor's single imputation method and their
`adjudication of reasons for dropout. However, 59 subjects (25%) in the UT-15C group
`did not have the week 12 walk test compared to 18 subjects (11%) in the placebo group.
`When the 59 UT-15C subjects are given worst rank, the p-value becomes 0.92. When the
`missing placebo subjects are assigned worst rank as well, the p-value becomes 0.21. It is
`of great concern that UT-15C had more than double dropout rate than placebo.
`
`14 (6%) subjects in the treprostinil group and 9 (8%) subjects in the placebo group died
`during the course of the study before the walking distance could be measured at Week 12
`(10 and 6 deaths respectively were listed as the reason for discontinuation but others died
`after discontinuation during 12 week period). Because of the 2:1 randomization, the
`percentage of deaths was approximately balanced between the two groups. There were an
`additional 45 (19%) subjects with missing data at week 12 in the treprostinil group and
`an additional 9 (8%) in the placebo group. Only 3 subjects total had missing data for the
`reason "In study, too ill to walk". 100% of the subjects that did not die and were not too
`ill to walk should have had a week 12 followup visit where the walking distance should
`have been measured and this value should have been used in the ITT analysis regardless
`of whether the subject took their randomized treatment.
`
`The primary analysis was complicated, but essentially all subjects were assigned a score
`between 0 and 1 based on their change from baseline walking distance. Higher scores
`indicate better change in walking distance. The average imputed score for the 59
`treprostinil subjects with missing Week 12 data is 0.36 while the average score for the 18
`placebo subjects is 0.11. From this, it is seen that there was a large amount of missing
`data and the way it was handled seemed to substantially favor showing a treatment effect
`(i.e., 0.36 vs. 0.11). Since it is preferable to make decisions based on observed data rather
`than made up data and there was a substantial amount of missing data in this study with
`twice as much missing in the treatment group, it seems worthwhile to consider other
`ways of handling the missing data that do not favor showing a treatment effect.
`
`
`Reference ID: 3201481
`
`
`
`Treprostinil tablets – NDA 203,496 Page 2 of 3
`⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯
`
`In this addendum, three additional analyses are considered:
`
`1. Analysis of change from baseline to Week 4 (earlier time point than used in the
`primary analysis) using other imputation methods; giving worst rank to all subjects
`(treprostinil and placebo) with missing data or only to all subjects in the trepostinil group.
`
`In the sponsor's ITT analysis of change from baseline to Week 4, the estimated placebo
`subtracted change from baseline is 14 m with a p-value of 0.0025. 25 subjects in the
`treprostinil group and 8 in the placebo group had missing value sat Week 4. In the
`sponsor's analysis, the imputed scores for these 25 subjects with missing values had a
`mean of 0.34 while the mean score for the 8 placebo subjects was 0.08. Again, the
`sponsor’s imputation seemed to substantially favor the treprostinil group.
`
`When these 25 subjects in the treprostinil group are all given the worst score, the point
`estimate of the placebo subtracted change from baseline is 10 m and the p-value is 0.063.
`If, in addition, the 8 placebo subjects are also given the worst score, then the p-value
`remains 0.063 (note: 7 of them already had the worst rank, so it only changes the rank for
`1 placebo subject).
`
`
`2. Analysis of change from baseline to Week 12 giving fewer subjects from treprostinil
`group the worst rank (i.e. not all 59 are given worst rank).
`
`When all 59 subjects in the treprostinil group are given the worst score, the p-value is
`0.92. If only the top 23 of these 59 are given the worst score and the remaining 36 scores
`are left "as is", the average score for all 59 treprostinil subjects with missing data is 0.13
`compared to an average score of 0.11 for the placebo subjects with missing data. The p-
`value for the analysis with this imputation is 0.051. 23 is the smallest number of subjects
`in the treprostinil group with missing data who were not already given the worst rank that
`would have to be given the worst rank to make the p-value above 0.05.
`
`The reasons given for missing data at Week 12 for these 23 subjects were: Adverse event
`(14), Consent withdrawn (2), Discontinued for other reasons (1), In study, unblinded or
`other (3), Lost to follow-up (3).
`
`3. Multiple imputation method for missing values at Week 12.
`
`Another approach is to use multiple imputation to impute the missing data. For each
`imputed dataset, the worst rank was given to subjects who died as was done in the
`original single imputation analysis. For subjects with missing value who did not die, a
`random score was chosen uniformly between 0 and 0.25; this is based on the concept that
`anyone with missing data would have fallen in the lowest quartile had they been coerced
`into actually doing the walk test at Week 12. Note that there are 45 subjects in the
`treatment group who have missing data at Week 12 for reason other than death and 9 in
`the placebo group. The average score in the sponsor's analysis for the 45 subjects was
`0.45 and the average score for the 9 placebo subjects was 0.16. From this imputed
`
`Reference ID: 3201481
`
`
`
`Treprostinil tablets – NDA 203,496 Page 3 of 3
`⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯
`
`dataset, the stratified treatment effect was estimated and then combined using a multiple
`imputation formula (Rubin, D.B. (1987) Multiple Imputation for Nonresponse in Surveys.
`J. Wiley & Sons, New York.). The multiple imputation analysis will have two important
`differences from the original single imputation. First, the imputed scores will have the
`same average value in both groups (namely, 0.125). Second, the uncertainty in imputing a
`value will be reflected in the variance. This uncertainty is not factored in when a single
`imputation is used and treated like a fixed value. The result of the multiple imputation
`analysis done this way is a p-value of 0.056. However, the validity of multiple imputation
`analysis relies on model assumptions including the assumption that the values are
`missing at random. This assumption implies that the large imbalance in the rate of
`missing data between the two treatment groups does not matter in statistical analysis,
`which can be very problematic in my view.
`
`There are other ways of imputing the missing values as part of the multiple imputation
`analysis. In general, those imputation schemes that tend to give better values to subjects
`with missing data will tend to favor showing a treatment effect because the number of
`subjects affected by these better imputed values is higher in the treprostinil group
`compared to the placebo group and vice versa. For example, if the missing data from the
`patients who did not die or too ill to walk are imputed by a random score smaller than
`0.125 on average, then the treatment difference will likely not be statistically significant.
`
`In summary, the robustness of the efficacy results depends heavily on how the missing
`data are treated in statistical analysis; the p-value can 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 the treprostrinil tablets has not been
`convincingly demonstrated, based on this study.
`
`
`Reference ID: 3201481
`
`
`
`---------------------------------------------------------------------------------------------------------
`This is a representation of an electronic record that was signed
`electronically and this page is the manifestation of the electronic
`signature.
`---------------------------------------------------------------------------------------------------------
`/s/
`----------------------------------------------------
`
`JOHN P LAWRENCE
`10/10/2012
`
`HSIEN MING J HUNG
`10/10/2012
`
`MARYANN GORDON
`10/10/2012
`
`Reference ID: 3201481
`
`
`
`
`
`
`
`U.S. Department of Health and Human Services
`Food and Drug Administration
`Center for Drug Evaluation and Research
`Office of Translational Science
`Office of Biostatistics
`
`
`
`Statistical Review and Evaluation
` CARCINOGENICITY STUDY
`
`NDA 203-496
`UT-15C (Treprostinil Diethanolamine)
`
`26 Week Carcinogenicity Study in Tg.rasH2 mice
`Sponsor: United Therapeutics Corporation
`55 T. W. Alexander Drive, P.O. Box 14186
`Research Triangle Park, NC 27709
`
`
`
`
`
`Electronic report submission: Dated Dec. 27, 2011 Electronic
`data submission: Dated January 31, 2012
`Standard
`
`Division of Biometrics -6
`Mohammad Atiar Rahman, Ph.D.
`Karl Lin, Ph.D.
`
`Division of Cardiovascular and Renal Products
`Xavier Joseph, Ph.D.
`Dan Brum
`
`
`Carcinogenicity, Dose response
`
`IND/NDA Number:
`Drug Name:
`Indication(s):
`Applicant:
`
`Documents Reviewed:
`
`Review Priority:
`
`Biometrics Division:
`Statistical Reviewer:
`Concurring Reviewer:
`
`Medical Division:
`Reviewing Pharmacologist:
`Project Manager:
`
`Keywords:
`
`
`Reference ID: 3138330
`
`(b) (4)
`
`
`
`NDA 203-496 UT-15C (Treprostinil Diethanolamine) Page 2 of 14
`
`
`
`
`
`
`Table of Contents
`
`1......................................................................................................................................... Background
`
`2................................................................................................................................................ Design
`2.1.
`Sponsor's analyses.............................................................................................................................3
`2.1.1. Survival analysis.................................................................................................................3
`2.1.2. Tumor data analysis............................................................................................................3
`Reviewer's analyses ..........................................................................................................................4
`2.2.1. Survival analysis.................................................................................................................4
`2.2.2. Tumor data analysis............................................................................................................4
`
`2.2.
`
`3..........................................................................................Evaluation of validity of the design of the study
`
`4.............................................................................................................................................Summary
`
`5............................................................................................................................................ Appendix
`
`3
`
`3
`
`5
`
`6
`
`7
`
`References ...................................................................................................................................................................14
`
`
`Reference ID: 3138330
`
`
`
` NDA 203-496 UT-15C (Treprostinil Diethanolamine) Page 3 of 14
`1. Background
`
`
`
`
`
`
`
`In this submission the sponsor included a report of an animal carcinogenicity study in mice. This study was
`intended to assess the carcinogenic potential of UT-15C (Treprostinil Diethanolamine) in Tg.rasH2 mice when
`administered orally via gavage at appropriate drug levels for about 26 weeks. All surviving animals were sacrificed
`at Week 27. Results of this review have been discussed with the reviewing pharmacologist Dr. Joseph.
`
`In this review the phrase "dose response relationship" refers to the linear component of the effect of treatment,
`and not necessarily to a strictly increasing or decreasing mortality or tumor incidence rate as dose increases.
`
`
`2. Design
`
`
`Two separate experiments were conducted, one in male and one in female mice. In each of these two
`experiments there were three treated groups, one vehicle control group (negative control), and one positive
`control group. A total of one hundred Tg.rasH2 mice of each sex were randomly assigned to treated and
`vehicle control groups in equal size of 25 animals per group. The positive control group had 15 mice in each
`sex. The dose levels for treated groups were 5.0, 10.0, and 20.0 mg/kg/day for male mice and 3.0, 7.5, and
`15.0 mg/kg/day for female mice. In this review the three treated groups are referred to as the low, medium,
`and high dose groups, respectively. The animals in the positive control group received a total of 3 urethane
`intraperitoneal injections on Study Days 1, 3, and 5. The animals in the vehicle control group received vehicle
`(sterile water for injection).
`
`All animals were observed twice daily for morbidity and mortality. They were also observed daily for clinical
`signs of toxicity. A detailed hands-on examination was performed on all animals once a week. Body weights
`for individual animals were measured once weekly beginning on first day of dosing through Week 13 and
`biweekly thereafter.
`
`
`2.1.
`
`Sponsor's analyses
`
`Survival analysis
`
`
`2.1.1.
`
`The sponsor presented a summary table of the mortalities of animals by sex. In the original protocol the
`sponsor proposed to analyze the mortality data using the Generalized Wilcoxon test. However, the submitted
`sponsor’s reports do not contain results of any formal statistical analysis of mortality data.
`
`Sponsor’s findings: The sponsor analysis showed one male death in medium dose group, one female death in
`medium dose group and one female death in high dose group. Besides, one female mouse from the low dose
`group and one female mouse from the medium dose group were sacrificed in moribund condition. The sponsor
`concluded that there was no significant differences in mortality among treatment groups in either sex.
`
`
`2.1.2. Tumor data analysis
`
` historical
`The sponsor presented a summary table of the tumor findings by sex, including the
`control ranges. The tumor data were analyzed using the method proposed by Peto et al. (1980), incorporating
`the context of observation. The positive control was compared to the vehicle control using the one-sided
`Fisher’s exact test.
`
`Adjustment for multiple testing: No adjustment for multiple testing was performed.
`
`Reference ID: 3138330
`
`(b) (4)
`
`
`
`
`
`
`
` NDA 203-496 UT-15C (Treprostinil Diethanolamine) Page 4 of 14
`Sponsor’s findings: The sponsor’s analysis did not show statistically significant dose response relationship
`among the treatment groups or increased incidence in the treated groups in any of the observed tumor types.
`The sponsor concluded that the incidences of all observed lesions were low and were within the historical
`control ranges established at
` laboratories, and the treatment by the test article did not increase the
`incidence of any neoplastic lesions.
`
`The sponsor’s analysis further showed statistically significant increased (p < 0.05) incidences of pulmonary
`tumors and splenic hemangiosarcomas in the positive control male and female mice when compared to their
`respective vehicle control.
`
`
`2.2. Reviewer's analyses
`
`Survival analysis
`
`
`To verify the sponsor’s analyses and conduct additional analysis suggested by the reviewing pharmacologist, this
`reviewer independently performed survival and tumor data analyses. Data used in this reviewer's analyses were
`provided by the sponsor electronically.
`
`2.2.1.
`
`The survival distributions of animals in all four treatment groups (vehicle control, low, medium, and high dose
`groups) were estimated by the Kaplan-Meier product limit method. The dose response relationship was tested
`using the likelihood ratio test and the homogeneity of survival distributions was tested using the log-rank test. The
`intercurrent mortality data of all treatment groups are given in Tables 1A and 1B in the appendix for male and
`female mice, respectively. The Kaplan-Meier curves for survival rates of all treatment groups are given in Figures
`1A and 1B in the appendix for male and female mice, respectively. Results of the tests for dose response
`relationship and homogeneity of survivals, are given in Tables 2A and 2B in the appendix for male and female
`mice, respectively.
`
`Reviewer’s findings: This reviewer’s analysis showed 0, 0, 1, and 0 deaths of male mice in vehicle control,
`low, medium, and high groups, respectively before the scheduled sacrifice on Week 27. In positive control
`group 7 male mice had natural death before Week 16 and the remaining 8 male mice were sacrificed on Week
`16 as the part of part of a planned interim sacrifice. Similarly there were 0, 1, 2, and 1 deaths of female mice in
`vehicle control, low, medium, and high groups, respectively before the scheduled sacrifice on Week 27. In
`positive control group 4 female mice had natural death before Week 16 and the remaining 11 female mice
`were sacrificed on Week 16 as the part of part of a planned interim sacrifice. This reviewer’s analysis did not
`show statistically significant dose response relationship in mortality across vehicle control, low, medium, and high
`dose groups in either sex. The pairwise comparisons also did not show statistically significant increased mortality in
`the low, medium, and high dose groups compared to the vehicle control group in either sex.
`
`Reviewer’s comment: The sponsor’s count showed a total of 3 deaths (1 male and 2 females), while this reviewer’s count showed a
`total of 5 deaths (1 male and 4 females). These discrepancies are due to the fact that there were two female mice, one in low dose group
`and one in medium dose group, that were killed by the sponsor in their moribund conditions. In the submitted data sets these animals
`were coded as naturally dead, which is reflected in this reviewer’s count.
`
`2.2.2. Tumor data analysis
`
`The tumor data were analyzed for dose response relationships and pairwise comparisons of vehicle control group
`with each of the treated groups. Both the dose response relationship tests and pairwise comparisons were
`performed using the Poly-k method described in the paper of Bailer and Portier (1988) and Bieler and Williams
`
`Reference ID: 3138330
`
`(b) (4)
`
`
`
`<1. The adjusted group size is defined as N*=Σ hs . As
`
` NDA 203-496 UT-15C (Treprostinil Diethanolamine) Page 5 of 14
`(1993). In this method an animal that lives the full study period ( maxw ) or dies before the terminal sacrifice but
`develops the tumor type being tested gets a score of hs =1. An animal that dies at week
`hw without a tumor
`k
`w ⎟
`w
`max
`an interpretation, an animal with score hs =1 can be considered as a whole animal while an animal with score
`hs <1 can be considered as a partial animal. The adjusted group size N* is equal to N (the original group size) if all
`animals live up to the end of the study or if each animal that dies before the terminal sacrifice develops at least one
`tumor being tested, otherwise the adjusted group size is less than N. These adjusted group sizes are then used for
`the dose response relationship (or the pairwise) tests using the Cochran-Armitage test. One critical point for Poly-k
`test is the choice of the appropriate value of k, which depends on the tumor incidence pattern with the increased
`dose. For standard rat and mouse studies, a value of k=3 is suggested in the literature. Hence, this reviewer used
`k=3 for the analysis of this data. For the calculation of p-values the exact permutation method was used. The
`tumor rates and the p-values of the tested tumor types are listed in Tables 3A and 3B in the appendix for male and
`female mice, respectively.
`
`Multiple testing adjustment: For the adjustment of multiple testing of dose response relationship in 104
`week mouse and rat studies, the FDA guidance for the carcinogenicity study design and data analysis suggests
`the use of test levels of α=0.005 for common tumors and α=0.025 for rare tumors for a submission with two
`species, and a significance level of α=0.01 for common tumors and α=0.05 for rare tumors for a submission
`with one species study in order to keep the false-positive rate at the nominal level of approximately 10%. A rare
`tumor is defined as one in which the published spontaneous tumor rate is less than 1%. For multiple pairwise
`comparisons of treated group with control the FDA guidance suggests the use of test levels α=0.01 for
`common tumors and α=0.05 for rare tumors, in order to keep the false-positive rate at the nominal level of
`approximately 10% for both submissions with two or one species.
`
`Since the present study is a 26 week study these rules are not applicable for the adjustment for multiple
`testing. With a conservative approach, in this reviewer’s analysis all p-values were compared against α=0.05.
`
`Reviewer’s findings: Using the test level of α=0.05, this reviewer’s analysis did not show statistically significant
`dose response relationship across vehicle control, low, medium , and high dose groups in the incidence of any of
`the observed tumor types in either sex. The pairwise comparisons also did not show statistically significant
`increased incidence of any of the observed tumor types in the low, medium, and high dose groups compared to
`the vehicle control group in either sex.
`
`
`3. Evaluation of validity of the design of the study
`
`
`Since, the tumor data analyses did not show statistically significant dose-response relationship or pairwise
`comparison in any of the tested tumor types in either sex, it is important to look into the validity of the design. For
`a transgenic mouse study using a positive control group, it is important to verify the performance of the positive
`control for the validation of the study. For a valid study the animals in the positive control group are expected to
`show significantly higher tumorogenicity compared to the animals in groups treated with the study compound
`group. Tables 4A and 4B show the results of the dose response relationship tests and the pairwise comparisons of
`tumor incidences using the data from positive control, low, medium, and high dose groups. The results show that
`the positive control group had statistically significant increased incidences of lungs adenoma, carcinoma,
`hemangiosarcoma and spleen hemangiosarcoma in both sexes of mice. The positive control group also had
`
`Reference ID: 3138330
`
`
`
`
`
`⎠⎞
`
`h
`
`⎜⎝⎛
`
`before the end of the study gets a score of hs =
`
`
`
`
`
` NDA 203-496 UT-15C (Treprostinil Diethanolamine) Page 6 of 14
`statistically significant increased mortality compared to the treated groups.
`
`The results indicate that the design of the study might be valid. However, other biological and toxic effects
`must be taken into consideration for the final evaluation.
`
`
`
`
`4. Summary
`
`
`In this submission the sponsor included a report of an animal carcinogenicity study in mice. This study was
`intended to assess the carcinogenic potential of UT-15C (Treprostinil Diethanolamine) in Tg.rasH2 mice when
`administered orally via gavage at appropriate drug levels for about 26 weeks.
`
`In this review the phrase "dose response relationship" refers to the linear component of the effect of treatment,
`and not necessarily to a strictly increasing or decreasing mortality or tumor incidence rate as dose increases.
`
`
`Design: Two separate experiments were conducted, one in male and one in female mice. In each of these
`two experiments there were three treated groups, one vehicle control group (negative control), and one
`positive control group. A total of one hundred Tg.rasH2 mice of each sex were randomly assigned to treated
`and vehicle control groups in equal size of 25 animals per group. The positive control group had 15 mice in
`each sex. The dose levels for treated groups were 5.0, 10.0, and 20.0 mg/kg/day for male mice and 3.0, 7.5,
`and 15.0 mg/kg/day for female mice. The animals in the positive control group received a total of 3 urethane
`intraperitoneal injections on Study Days 1, 3, and 5. The animals in the vehicle control group received vehicle
`(sterile water for injection).
`
`Results: The tests did not show statistically significant dose response relationship in mortality across vehicle
`control, low, medium, and high dose groups in either sex. The pairwise comparisons also did not show statistically
`significant increased mortality in the low, medium, and high dose groups compared to the vehicle control group
`in either sex.
`
`The tests did not show statistically significant dose response relationship across vehicle control, low, medium, and
`high dose groups in the incidence of any of the observed tumor types in either sex. The pairwise comparisons also
`did not show statistically significant increased incidence of any of the observed tumor types in the low, medium,
`and high dose groups compared to the vehicle control group in either sex. The positive control group showed
`statistically significant increased mortality and incidences of lungs adenoma, carcinoma, hemangiosarcoma, and
`spleen hemangiosarcoma in both sexes of mice.
`
` Mohammad Atiar Rahman, Ph.D.
` Mathematical Statistician
`Concur: Karl Lin, Ph.D.
` Team Leader, Biometrics-6
`
`cc:
`Archival NDA 203-496
`
`Dr. Joseph Dr. Machado
`Mr. Brum Dr. Lin
` Dr. Rahman
` MS. Patrician
`
`Reference ID: 3138330
`
`
`
`
`
` NDA 203-496 UT-15C (Treprostinil Diethanolamine) Page 7 of 14
`5. Appendix
`
`Table 1A: Intercurrent Mortality Rate
`Male Mice
`
`
`
`
`
` _Veh. Control_ _____Low_____ ____Medium___ _____High_____ _Pos. Control_
`
` No. of No. of No. of No. of No. of
`
` Week Death Cum. % Death Cum. % Death Cum. % Death Cum. % Death Cum. %
`
` ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
`
` Week 0 - 1 . . . . . . . . 3 20
`
` Week 11-20 . . . . . . . . 4 47
`
` Week 21-26 . . . . 1 4 . . . .
`
` Ter. Sac. 25 100 25 100 24 96 25 100 8* 53
`
` ------------------------------------------------------------------------------------------------
`
` Total N=25 N=25 N=25 N=25 N=15
`
`
`* Animals in positive control were sacrificed on week 16
`
`
`Table 1B: Intercurrent Mortality Rate
`Female Mice
`
`
`
`
`
` _Veh. Control_ _____Low_____ ____Medium___ _____High_____ _Pos. Control_
`
` No. of No. of No. of No. of No. of
`
` Week Death Cum. % Death Cum. % Death Cum. % Death Cum. % Death Cum. %
`
` ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
`
` Week 0 - 1 . . . . 1 4 . . 1 7
`
` Week 11-20 . . 1 4 1 4 . . 3 27
`
` Week 21-26 . . . . . . 1 4 . .
`
` Ter. Sac. 25 100 24 96 23 92 24 96 11* 73
`
` ------------------------------------------------------------------------------------------------
`
` Total N=25 N=25 N=25 N=25 N=15
`
`
`* Animals in positive control were sacrificed on week 16
`
`
`
`
`
`
`Table 2A: Intercurrent Mortality Comparison
`Male Mice
`
`
` Test Statistic P_Value*
` ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
` Dose-Response Likelihood Ratio 0.9866
` Homogeneity Log-Rank 0.3916
`
` * The p-values were calculated using data from the vehicle control, low, medium and high dose groups
`
`Table 2B: Intercurrent Mortality Comparison
`Female Mice
`
`
`
`
` Test Statistic P_Value*
` ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
` Dose-Response Likelihood Ratio 0.8950
` Homogeneity Log-Rank 0.5496
`
` * The p-values were calculated using data from the vehicle control, low, medium and high dose groups
`
`Reference ID: 3138330
`
`
`
`
`
` NDA 203-496 UT-15C (Treprostinil Diethanolamine) Page 8 of 14
`
`Table 3A: Tumor Rates and P-Values for Dose Response Relationship and Pairwise Comparisons
`of Treated Groups with Vehicle Control in Male Mice
`
`
`
` 0 mg 5 mg 10 mg 20 mg ____________P-Value_______________
`
` Veh. Cont. Low Med High Dose
`
` Organ Name Tumor Name N=25 N=25 N=25 N=25 Resp VCvs L VCvs.M VCvs.H
`
` ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
`
` cavity, nasal adenocarcinoma 0 0 1 1 0.1869 . 0.5000 0.5000
`
`
`
` ear papilloma 0 0 1 0 0.5000 . 0.5000 .
`
`
`
` harderian glands adenoma 0 0 1 2 0.0606 . 0.5000 0.2449
`
`
`
` lungs with bronchi alveolar-bronchiolar adenoma 2 4 1 1 0.8020 0.3336 0.5000 0.5000
`
`
`
` lungs with bronchi adenoma+carcinoma 2 4 1 1 0.8020 0.3336 0.5000 0.5000
`
`
`
` lung+spleen+testes hemangiosarcoma 0 3 3 1 0.4643 0.1173 0.1173 0.5000
`
`
`
` sple