CENTER FOR DRUG EVALUATION AND
`
`RESEARCH
`
`APPLICA TION NUMBER:
`
`22-301
`
`STATISTICAL REVIEW! S2
`
`

`

`Office of Biostatistics
`
`US. Department of Health and Human Services
`Food and Drug Administration
`Center for Drug Evaluation and Research
`Office ofTranslational Sciences
`
`STATISTICAL REVIEW AND EVALUATION
`
`CLINICAL STUDIES
`
`NDA/Serial Nnmber:
`
`22—301
`
`Drug Name:
`
`Encapsulated Mesalamine Granules, 0.375 g (eMG)
`
`(S-aminosalicylic acid or 5—ASA)
`
`Indication(s):
`
`Maintenance of Remission of Ulcerative Colitis
`
`Applicant:
`
`Date(s):
`
`Salix Pharmaceuticals
`
`Stamp: December 21, 2007; PDUFA October 31, 2008
`
`Review Priority:
`
`Standard
`
`'
`
`Biometrics Division:
`
`Division of Biometrics 111
`
`V Statistical Reviewer:
`
`Shahla S. Farr, M.S.
`
`Concurring Reviewer:
`
`Mike Welch, Ph.D., Deputy Din, DBIII
`
`Medical Division:
`
`Division of Gastroenterology Products
`
`Clinical Team:
`
`Aisha Peterson, M.D., John Hyde, M.D. (TL)
`
`Project Manager:
`
`Cristi Stark, M.S.
`
`Keywords: NDA Review, Clinical studies, Multiple Comparisons, Hierarchical Analysis
`
`

`

`Table of Contents
`
`1.
`
`EXECUTIVE SUMMARY .................................................................3
`
`1.1
`1.2
`1.3
`
`CONCLUSIONS AND RECOMMENDATIONS ............................... 3
`BRIEF OVERVIEW OF CLINICAL STUDY ................................... 3
`STATISTICAL ISSUES AND FINIDINGS ...................................... 3
`
`2.
`
`INTRODUCTION........................................................................... 4
`
`2.1
`2.2
`
`OVERVIEW ...........................................................................4
`DATA SOURCES .................................................................... 5
`
`3.
`
`STATISTICAL EVALUATION.......................................................... 5
`
`3.1
`3.2
`
`EVALUATION OF EFFICACY ................................................... 5
`EFFICACY RESULTS .......................................................' ....... 8
`
`4.
`
`FINDINGS IN SPECIAL/SUBGROUP POPULATIONS .......................... 13
`
`4.1
`
`GENDER, RACE, AGE AND OTHER SPECIAL/
`SUBGROUP POPULATIONS .................................................... l3
`
`5.
`
`SUMMARY AND CONCLUSIONS .................................................... 15
`
`‘ 5.1
`5.2
`
`STATISTICAL ISSUES AND COLLECTIVE EVIDENCE ................. 15
`CONCLUSIONS AND RECOMMENDATIONS .............................. 15
`
`APPENDIX .............................................................. 16
`
`

`

`1 EXECUTIVE SUMMARY
`
`1.1 Conclusions and Recommendations
`
`Based on data from two randomized and controlled studies, Encapsulated Mesalamine Granules,
`0.375 g (eMG) (5-aminosalicylic'acid or S-ASA) taken once a day for six months appear to be
`efficacious for maintenance of remission of Ulcerative Colitis, as assessed by relapse-free rate,
`the primary endpoint. The results from Study MCUP3003 provided clearer evidence of efficacy
`compared to StudyMCUP 3004. Both studies failed to show consistent secondary endpoint
`efficacy. For Study MCUP3 004, the trial completed at approximately the same time the protocol
`was amended to reduce the sample size; thus interpretation of even the primary result is
`problematic, and, at best, that study should be considered as showing marginal efficacy in
`support of the first trial.
`
`1.2 Brief Overview of Clinical Studies
`
`The sponsor has submitted results from two, Phase 3, double-blind, randomized, parallel-group,
`placebo-controlled, multi-center trials to investigate the efficacy and safety of Encapsulated
`Mesalamine (S—aminosalicylic acid or 5-ASA), Granules (eMG) for maintenance of remission of
`Ulcerative Colitis (U_C). The studies were conducted in males and non-pregnant females ages 18
`years and older, with mildly to moderately active UC who had previously demonstrated
`remission of UC. Study MPUC3003 enrolled 160 US. subjects and 145 in Russia; Study
`MPUC3004 enrolled 103 in the US. and 154 in Russia.
`
`The primary efficacy analysis endpoint was the proportion of subjects who were relapse-free
`after six months of treatment. Relapse or treatment failure was defined as a rectal bleeding score
`of l or more and a mucosal appearance score of 2 or more as described in the revised Sutherland
`Disease Activity Index (DAI). In addition, subjects who experienced a UC flare or initiated
`medication used'previously to treat UC were also considered treatment failures. Subjects were
`randomized 2:1 to receive the test product (eMG) or placebo. Secondary endpoints included
`rectal bleeding score, mucosa] appearance score, physicians’ rating of disease activity, and
`individual components of the Sutherland DAI.
`
`1.3 Statistical Issues and Findings
`For Study MPUC3003, the primary efficacy comparison shows a highly statistically significant
`difference between Mesalamine and placebo (p < .001). For Study MPUC3004, the primary
`efficacy results were also statistically significant (p = .029).
`
`The sponsor’s imputation strategy assigned treatment failures to subjects who terminated the
`study early but only if the reason for drop-out was related to lack of efficacy or a UC-related
`adverse event. This approach to handling drop-outs was changed in late-stage protocol
`amendments after study completion; the original plan was to consider all subjects who
`terminated early as treatment failures. The results from this more conservative analysis do not
`change the efficacy conclusions for the first study; however, the efficacy results for Study
`MPUC3004 are marginal (p = .046).
`
`

`

`To control for experiment-wise type I error, the sponsor planned a hierarchical testing strategy
`for the secondary endpoints, although the order of testing was specified in a protocol amendment
`afier completion of studies but prior to unblinding. For Study MPUC3003, only the first
`secondary endpoint (change from baseline in rectal bleeding score) showed a statistically
`significant improvement from baseline. None of the secondary endpoints were statistically
`significant in Study MPUC3004.
`
`The planned sample size for Study MPUC3 004 was reduced in a late—stage protocol amendment.
`A total of 257 patients instead of the planned 300 were analyzed. This could be interpreted as an
`unplanned or early stopping of the study. Even though the decision appears to have been made
`prior to breaking the blind, the study completion date precedes the protocol amendment date. A
`sensitivity analysis shows that if the additional 43 subjects were enrolled and had a 68%
`treatment success rate for both treatment groups (consistent with the observed placebo rate) then
`the primary ITT analysis would have failed (p = .06). This may suggest that the study was
`terminated early to avert possible failure of the primaryendpoint.
`
`The reviewer performed several subgroup analyses using the combined study data. These results
`suggest a smaller effect size for the Russian and males subpopulations but can be attributed to
`higher placebo response rates. Efficacy results are consistent across baseline disease severity as
`well as time on remission.
`'
`'
`
`2 INTRODUCTION
`
`2.1 Overview
`
`Encapsulated mesalamine (5—aminosalicylic acid or 5-ASA) granule is a capsule oral dosage
`form of mesalamine developed by Salix Pharmaceuticals, Inc., as a once daily (QD) dosing
`regimen in the maintenance of remission of ulcerative colitis (UC). The proposed dosage and
`
`administration is four 0.375 g eMG capsules (1.5 g/day) administered QD
`for
`subjects in remission of UC. Several mesalamine—containing dosage forms have been approved
`in the United States for use in UC over the last twenty years. However, there is, currently, no
`marketed mesalamine product in the US with QD dosing for the maintenance or remission of
`UC.
`'
`
`11(4)
`
`Disease activity in the two trials MPUC3003 and MPUC 3004 was measured using the revised
`Sutherland DAI. The Sutherland DAI was selected because it represents a historical standard for
`assessing the symptoms of UC (rectal bleeding, mucosa] appearance, physician’s rating, and
`'
`stool fi'equency). No standard scoring system for measuring UC disease activity has been
`validated for clinical use. At the request of the agency and to clarify the diagnosis of UC
`remission, two changes were made to the Sutherland definition of mucosal appearance for
`~ protocolsMPUC3003 and MPUC 3004 before the studies were started. For these studies, the
`term “mild friability” was removed from the mucosa] appearance score of 1, and the term
`“moderate friability” was removed from the mucosa] appearance score of 2 as defined in
`Sutherland, et al., 1987. (See end-of-phase 2 meeting minutes dated October 6, 2004.)
`
`The sponsor conducted two similar, Phase 3, double-blind, randomized, parallel-group, placebo-
`controlled, multi-center trials to investigate the efficacy and safety of Encapsulated Mesalamine
`
`4
`
`

`

`(S-aminosalicylic acid or 5-ASA) Granules (eMG) for the Maintenance of Remission of
`Ulcerative Colitis for the duration of 6 months (Study MPUC3003 and Study MPUC 3004).
`Subjects were randomized 2:1 to receive the test product (eMG) or placebo. A total of 305
`subjects (209, 68.5% in eMG and 96, 31.5% in placebo) from 45 centers took part in study
`MPUC3003, and a total of 257 patients (164, 63.8% in eMG and 93, 36.2% in placebo) from 37
`sites participated in study MPUC 3004. The primary analysis efficacy endpoint was the
`proportion of subjects who were relapse-free after 6 months of treatment, where relapse was
`defined by the revised Sutherland Disease Activity Index as a rectal bleeding score 2 l and a
`mucosa] appearance score 2 2.
`
`To be eligible for these studies, subjects were to have a historically confirmed diagnosis of UC in
`remission for at least 1 month and not more than 12 months, and a confirmed current remission
`of UC defined as a rectal bleeding score of 0 and a mucosa] appearance score of 0 or 1 using a
`revised Sutherland Disease Activity Index (DAI). Subjects were also required to have a history
`of at least one flare with symptoms within the past 1 to 12 months that required therapeutic
`intervention but not to have taken steroids or immunosuppressive agents within 30 days of
`screening.
`
`The sponsor also submitted interim safety data from an ongoing open-label extension safety
`study (MPUC3005) which includes subjects who participated in Study 3003 and 3004 as well as
`new subjects who did not participate in the pivotal studies. Refer to theMedical Officer’s review
`for the safety assessment.
`'
`
`Table l: Descri tion of the Efficac Studies
`
`
`
`
`
`
`# of Subjects
`'
`Study Description
`Study #
`Total ITT: 305
`1.5 g eMG or Placebo
`MPUC3003 Males and non-pregnant women
`
`
`eMG:
`209
`/QD
`in remission from UC for at least
`Placebo: 96
`1 month and not more than 12
`
`
`
`
`months (US and Russia)
`
` Total ITT: 257
`
`
`
`1.5 g eMG or Placebo
`Males and non-pregnant women
`eMG:
`164
`in remission from UC for at least
`/QD
`
`
`Placebo: 93
`1 month and not more than 12
`
`
`
`months (US and Russia
`
`
`
`
`MPUC3004
`
`
`
`In this review, for the sake of brevity, the two studies will be referred to as study 3003 or study
`3004.
`
`2.2 Data Sources
`
`This NDA was submitted in CTD (paper) format. However, the datasets were provided
`electronically and are located at: \\FDSWAI50\NONECTD\N22301\N_000\2007-l2-21
`
`

`

`3.
`
`STATISTICAL EVALUATION
`
`3.1 Evaluation of Efficacy
`Study Design:
`r
`The sponsor conducted two, Phase 3, double—blind, randomized, parallel-group, placebo-
`controlled, multi:center trials to investigate the efficacy and safety of Encapsulated Mesalamine
`(S-aminosalicylic acid or 5-ASA), Granules (eMG) for the Maintenance of Remission of
`Ulcerative Colitis (UC) for the duration of 6 months (StudyMPUC3003 and Study MPUC 3004)
`in males and non-pregnant females ages 18 and older with mildly to moderately active UC.
`Subjects were to have a confirmed diagnosis of UC in remission for at least 1 month and not
`more than 12 months, as discussed above.
`
`Subjects were randomized 2:1 to receive the test product, eMG or placebo once daily. A total of '
`305 subjects (209, 68.5% in eMG and 96, 31.5% in placebo) fi'om ‘45 centers took part in study
`MPUC3003 and a total of 257 patients (164, 63.8% in eMG and 93, 36.2% in placebo) from 37
`sites participated in study MPUC 3004. Study MPUC3003 enrolled 160 US. subjects and 145 in
`Russia; Study MPUC3004 enrolled 103 (US) and 154 (Russia).
`
`The primary efficacy analysis population is Intent-to-Treat (ITT), which is defined as all subjects
`who were randomized to the study and had taken at least one dose of the study medication.
`
`Primary Objective:
`. The primary objective of the pivotal studies was to evaluate the efficacy and safety of eMG
`compared to placebo in men and non—pregnant women for maintenance of remission from UC as
`measured by rectal bleeding and endoscopic mucosa] appearance after 6 months.
`
`-
`Primafl Endpoints:
`The primary analysis efficacy endpoint is the proportion of subjects who were relapse-flee after
`six months of treatment. Relapse or treatment failure was defined as a rectal bleeding score of 1
`or more and a mucosa] appearance score of 2 or more as described in the revised Sutherland
`Disease Activity Index (DAI), shown in Table 2.
`
`In addition, subjects who experienced a UC flare or initiated medication used previously to treat
`UC were also considered treatment failure.
`
`

`

`
`
`Table 2: TRevis Sutherland DiseaseActlvi
`
`
`
`Stool Frequency
`0 = Normal
`
`
`l_= l to 2 stools/day more than normal
`
`
`2 = 3 to 4 stools/day more than normal
`3 =
`sootIs/da more thanoal
`
`
`
`
`‘
`
`Appearance
`
`3 = Mostly blood
`
` I = Steaks of blood
`
`2 = Obvious blood
`
`
`
`
`0 = Intact mucosa with preserved or distorted
`:
`
`
`
`
`vessels
`_
`
`
`l = Erythema, decreased vascular pattern,
`I
`
`
`3 granularity, no mucosal hemorrhage
`
`
`
`‘
`2 = Mucosal hemorrhage without blood in the
`3 lumen or gross ulceration, marked erythema,
`3 absent vascular pattern, small ulcers
`
`
`‘
`3 = Blood in lumen, gross ulceration,
`
`
`
`
` flng
`of Disease Activity
`
`
`2 = Moderate
`
`3 = Severe
`
`
`
`
`
`“mild friability” from the
`rie to reoev the tre
`* for protocols 3003 and 3004 the Sutherland DAI
`mucosal appearance score of 1 and to remove the term “moderate friability’ from the mucosal appearance score of 2.
`
`Secondafl Endpoints:
`Seven secondary endpoints were defined in the protocol. Protocol amendment 02, dated July 16,
`2007 for Study 3003 and amendment 02 dated August 9, 2007 for study 3004, modified and
`ordered the secondary endpoints for hierarchical testing in order to control for the type I error. It
`should be noted that these amendments occurred after study enrollment completed but, according
`to the sponsor, prior to unblinding. The seven secondary efficacy endpoints, in hierarchical
`order are:
`
`l) Rectal Bleeding: The number and proportion of subjects in each level of change from
`baseline in rectal bleeding score at months 1, 3 and 6.
`2) Mucosal Appearance: The number and proportion of subjects in each level of change from
`baseline in mucosal appearance score at month 6.
`3) Physician's Rating Of Disease Activity
`4) The Southerland DAI Score 5 2 with no Individual Component >1 and a Rectal Bleeding
`Score of 0
`
`5) Mean Change from Baseline in the Revised Sutherland DAI Score at Month 6
`6) Relapse-Free Duration
`7) An Assessment of the Revised Sutherland DAI Score for Stool Frequency.
`
`

`

`If testing of the primary endpoints showed statistical significance, then testing was to be
`conducted for the secondary endpoints until a non-significant p-value was found at p>0.05.
`Once a non-significant result was obtained, the testing was to be stopped.
`
`For each study, the final statistical analysis plan (SAP) refers to the analysis of the primary
`endpoint using the Per-Protocol (PP) population as a secondary analysis. However, in their
`submission, the sponsor has presented the PP analyses as sensitivity analyses for the primary. In
`the SAP, this PP analysis was not considered part of the hierarchical testing.
`
`Randomization:
`
`Eligible subjects received a unique identifier by a consecutively assigned subject ID number by
`. the order of enrollment within each study center. Treatments were assigned randomly via a
`randomization schedule using a 2:1 (2 eMG : l Placebo) ratio.
`
`Sample Size Calculation:
`For each study, the sample size was based on the assumption that 70% of the mesalamine-treated
`subjects and 50% of the placebo-treated subjects would be relapse-free at the end of 6 month of
`treatment. Study 3003 was designed to provide at least 90% power (Beta= 0.1) to detect this
`treatment difference between eMG and placebo in the proportion of subjects who were relapse-
`free after 6 weeks using a 2—sided significance level of 5% (alpha= 0.05) and 2:1 allocation ratio.
`A total of 200 subjects were randomized to the eMG and 100 randomized to the placebo am.
`
`In Protocol Amendment 02, dated August 9, 2007, Study 3004 was revised to reduce the number ,
`of planned subjects from 300 to 250. Based on the sponsor’s calculations, a total of 250 subjects
`would provide at least 80% power (beta=0.20) to reject the null hypothesis of no difference
`between eMG and placebo. A total of 257 patients (164, 63.8% in eMG and 93, 36.2% in
`placebo) participated in study 3004. The sponsor states this amendment occurred prior to
`breaking the study blind, but the study completion date is shown as August 8, 2007.
`
`Statistical Methodology:
`Statistical testing of the primary endpoint was done using 2-sided Chi-square test with an alpha
`level of 0.05. A Cochran-Mantel-Haenszel (CMH) test, controlling for country was performed.
`For the categorical secondary endpoint variables a 2-sided Chi-square was used. For all
`continues variables an Analysis of Variance or Analysis of Covariance was used, appropriately.
`
`Analysis Population:
`The primary analyses were performed on the ITT population, which was defined as all subjects
`who were randomized to the study and had taken at least one dose of study medication.
`
`Handling of Missing Data
`Last observation carried forward (LOCF) was applied for subjects who terminated the study
`early, except that subjects who terminated due to lack of efficacy or UC—related adverse events
`were classified as treatment failures. After study completion but before unblinding, the sponsor
`held data reviews to determine reasons for early study terminations. This imputation method
`was changed from the original protocols which stipulated that all subjects who dropped out early
`would be considered treatment failures, regardless of the reason for early termination. This
`
`8
`
`

`

`change appeared in Amendment 2 dated July 16, 2007 for study 3003 and Amendment 2 dated
`August 9, 2007, for study 3004.
`
`As noted above, these protocol changes occurred after the studies were completed. This
`reviewer performed the primary analyses using both the original and modifed methods for
`handling dropouts. The results are shown in the next section, under “Analyses of the Primary
`Endpoint.”
`
`3.2 Efficacy Results
`Patient Disposition, Demographics and Baseline Characteristics:
`Subjects were randomized 2:1 to receive the test product (eMG) or placebo once daily in subjects
`who had demonstrated remission of UC. A total of, 305 subjects (209, 68.5% in eMG and 96,
`31.5% in placebo) from 45 centers took part in study MPUC3003, and a total of 257 patients
`(164, 63.8% in eMG and 93, 36.2% in placebo) fi'om 37 sites participated in study MPUC3 004.
`Table 3 shows the disposition of the subjects by study.
`
`Both studies were conducted concurrently. Study 3003 began on December 20, 2004 and
`completed on April 26, 2007. Study 3004 was initiated on December 24, 2004 and was
`completed on August 8, 2007.
`
`
`
`
`
`.22
`T_able3. Down ofSub'ects by Study (Reviewer’5 Table
`
`
`
`_____MMMDWNE;2,___.___m_2w_,
`
`
`E n=164
`
`n=931=257
`11—8/257=32%
`
`
`
`
`n-=-961=305
`n=209
`113 (5407 2930.2
`65/209-31%
`E_ 47
`Reason for Withdrawal
`
`
`
`AdverseEvent I24
`
`Lost to Follow—Up
`
`
`Lack of Efficacy
`Subject Request
`
`
`Noncompliance
`
`Other
`6
`
`
`E9
`E2
`2E5
`2
`l
`
`6
`3
`22
`l
`0
`
`4
`
`
`
`
`
`
`As it is shown in Table 3, most drop-outs occurred as a result of adVerse events and/or lack of
`efficacy. In both studies, drop-out rates were similar, although the eMG groups had slightly
`higher drop--out rates compared to the placebo groups.
`
`Table 4 presents the demographics and baseline characteristics of the subjects by study.
`
`

`

`Table 4: Demora nhics and Baseline Characteristics of Sub'ects b Stud
`
`
`
`_
`
`Se
`Male
`Female
`
`
`_1 Study 3003
`Study 3004
`1
`
`' __J
`F: ,
`
`
`11Mesalam1ne mPa—o— Total
`11 Mesalarm”_
`_tal“
`11
`(F209)
`1
`(0:96)
`i
`(N=305)
`(F154)
`1
`1
`(N
`
`
`
`
`’
`1
`‘
`1
`1
`/2
`41-2/209=44% 53/96=55% 145/305=48% 1 4/164=45% ‘8/93=52% 122 57=47%
`
`
`1°11]7/209=563/96=45% 160/305=52%',»0/l64=55%1‘5/93=48%1135/257=53%
`
`
`
`1
`1
`‘
`1
`1
`1
`
`
`
`
`
`131/257=90%
`
`
`1; 6/257=10%
`1
`a
`5
`‘
`1
`1
`3
`.
`
`
`1°l88/209=9085/96=89% 1'73/3o5=90% 156/163=96% |0/93=97% ‘y46/256=96%
`11
`
`1r1/209=10%111/96=]1% 2/305=10% .
`/l63=4%
`3/93=3%
`10/256=4%
`
`
`
`
`
`1 1 1
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`: aseline Disease *
`Severity Category
`1
`
`Nomal or None
`110/257=43%
`11 1/164=43% 1
`
`2 1
`11: 3/164=57% ‘54/93=58% 147/257=57%
`
`I'aseline Remission Dur.
`1
`1
`1
`11
`1
`1
`s 13 Weeks
`1110/209=539/96=51% ‘159/305=52% 4/164=57% 14/93=47% 138/257=54%
`
`> 13 Weeks
`19/209=47% ‘7/96=49% 146/305=48%1;0/164=43% 9/93=53%1119/257=46%
`
`'-‘-‘1-"-%-‘-
`Mean (Std)
`6.8 (5.5)
`6.4 (4.8)
`i
`116(4.6)
`6(4.4)
`;
`
`110/209=53°150/96=52% 160/305=52% 2/164=38% 1/93=44% 103/257=40%
`US
`
`
`l'9/209=47%16/96=48%1145/305=48%'102/164=62%52/93=S6% .154/257=60%
`Russia
`
`
`
`
`*Disease Severity: Sum of all DAI scores, ranging from 0 (normal) to 12 (maximum score).
`
`In both studies, the distribution of subjects in both treatment groups were balanced with respect
`to demographics and baseline characteristics, such as gender, age and age category, race (White
`vs. non-Whites), baseline disease severity, baseline remission duration (5 13 Weeks vs. > 13
`Weeks), baseline BMI and country (US. vs. Russia).
`
`Analyses of the Primary Endpoints:
`In the original protocols, the population for the primary analyses were to include all subjects who
`discontinued the study early counted as relapse or ‘failure’. However, the sponsor changed this
`definition in protocol amendments dated after the completion of the studies, and assigned relapse
`to early terminators only if they dropped out due to lack of efficacy or UC-related adverse
`events.
`
`Primary analyses using both methods of imputation were done by the reviewer. Table 5 shows
`the sponsor’s as well as the reviewer’s efficacy analysis results of the primary endpoint variable.
`
`10
`
`

`

`Table 5:Prima Efficac Endoint"
`
`
`
`
`
`
`Mesalamine
`
`Placebo
`
`
`
`
`
`
`P-Value
`
`95% CI for
`
`Difference
`Stud 3003
`.001
`21°13, 32%)
`56/96=58%
`165/209=°
`Re1ase (S_ITT)**
`55/93=59% _ <0.001
`j 157/200=78.5%
`No Relase (PP
`
`
`N0 Rela'S‘éwlTT *i14329‘4/063-% 49/,51.
`.-,1-7°,,,,’229-)
`<0001..-__
`_ —;_,_
`
`
`
`___*77fim _ _.. ____
`ia¥flga 1‘ _,,_‘,.___._ __.___.___.—.—
`-:_;_
`63/93=68°
`j 12% (1.1%, 24%) 3—
`131/164=80%.
`No Relapse (S_ITT)**
`58/86=67% 1-11 0.027
`129/161=80%
`No Relapse (PP)
`
`55/93=59%
`1 12% (0%, 24.5%)
`1 0.046
`No Rela-se(RITT *** 3 117/164=71%
`* The primary analysis efficacy endpoint is the proportion of subjects who were relapse-free afier 6 months of
`treatment. Relapse or treatment failure was defined as a rectal bleeding score of 1 or more and a mucosal
`appearance score of 2 or more as described in the revised Sutherland Disease Activity Index (DAI).
`**Sponscr’s ITT analysis (early dropouts as relapse only lack of efficacy or if UC—related AE occurred.
`***Reviewer’s ITT analysis, all early withdrawals as relapse.
`
`The primary efficacy endpoint comparisons based on sponsor’s revised imputation strategy
`shows a highly statistically significant difference between Mesalamine and placebo in study
`3003 (p<0.001). Study 3004, also, demonstrated statistically significant results (p=0.029).
`When these analyses were repeated using per-protocol population the results were comparable to
`that of the intent-to-treat population. The reviewer’s findings were consistent with the sponsor’s
`results.
`
`Based on the more conservative approach where all early drop-outs are classified as having
`relapse, the overall efficacy conclusions for the ITT analyses are not changed; however, the
`results for study 3004 are only marginally significant at p = .046. (Based on Fisher’s exact test,
`p = .054.)
`
`This reviewer performed additional analyses of the primary endpoint, each analysis adjusting for
`a single factor [(country, site, gender, age group (<65 or 265), race (White/Non-White) and
`baseline severity category (0, Z 1)] using the CMH Chi-square test. For Study 3003, each
`analysis showed highly significant differences between treatment groups (p < 0.001). For Study
`3004, these same analyses resulted in statistical significance (0.03 < p < 0.046).
`
`Analyses of the Secondary Endpoints:
`The sponsor analyzed the secondary endpoints using the ITT population with a LOCF imputation
`method. The sponsor had specified in the SAP’s that if statistical significance was not achieved
`for any of the secondary endpoints, all subsequent secondary endpoints would be considered
`exploratory in nature. For study 3003, the second secondary endpoint (mucosal appearance)
`failed testing, and for study 3004, the first secondary endpoint (rectal bleeding) failed testing.
`
`Table 6 presents the reviewer’s analysis of the first two secondary endpoints. The reviewer’s
`analysis method for the endpoint is based on a Chi-square test; however results are consistent
`with those of the sponsor.
`In both studies for both the treatment arms, the majority of subjects
`
`11
`
`

`

`had no change from baseline for either bleeding or mucosa] appearance. This is noted in the
`table in bold font. The sponsor’s results for the other secondary endpoints, Physician’s rating
`and the Sutherland DAI component scores are shown in the Appendix.
`
`Table 6: Secondary Efficacy Endpoint (Reviewer’s Results)
`
`P-Value
`Placebo
`Mesalamine
`Study 3003
`__:_____,.,,,,.Ww_1.-.____._._____ __m
`, Rectal Bleeding n (%)
`0.01*
`-l
`
`
`
`
`1 (1%)
`64 (67%)
`11 (11%)
`19 (20%)
`1 (1%)
`
`0 (0%)
`170 (81%)
`22 (10.5%)
`16 (8%)
`
` 32 (15%)
`
`13 (13.5%)
`129 (62%) ’
`51 (53%)
`32 (15%)
`20 (21%)
`
`14 (7%)
`11 (11%)
`2 (1%)
`1 (1%)
`
`*SPOHSOY’S p’Value=0.008, using CMI—l controlling for countryVVTTWW
`r
`“Sponsor’s p—value=0.098, using CMH controlling for country
`
`Table 6, Cont’d: Secondary Efficacy Endpoint (Reviewer’s Results)
`
`Chane from Baseline
`
`n=164
`
`Rectal Bleeding n (%)
`
`3
`
`1 (0.6%)
`138 (84%)
`V 12 (7%)
`12 (7%)
`
`Mucosal Appearance n (%)
`-1
`0
`1
`2
`
`,
`
`26 (15.9%)
`104 (63.4%)
`18 (11.0%)
`16 (9.8%)
`
`l (1%)
`69 (74%)
`13 (14%)
`9_(10%)
`(1%)
`
`~
`13 (14.0%)
`56 (60.2%)
`14 (15.1%)
`10 (10.8%)
`
`For Study 3003, based on the hierarchical testing, only rectal bleeding showed a statistical
`significant result. In Study 3004, rectal bleeding did not show a statistically significant
`difference between eMG and placebo. Therefore, all secondary results for study 3004 should be
`considered exploratory.
`
`l2
`
`

`

`.
`Additional Analyses:
`In consultation with the clinical reviewer, I conducted additional analyses of bleeding and
`mucosal appearance status at the end of the study as assessed by binary analysis. A rectal
`bleeding score of 0 is defined as “No Bleeding”. A mucosa] appearance score of 0 was
`considered as “Normal”. Tables 7 and 8 show the results of these exploratory analyses by study.
`
`
`
`
`
`Table 7: Bleedin-
`
`Study 3003
`
`Bleeding
`Yes
`No
`Study 3004
`
`'
`
`Mesalamine
`(n=209)
`
`64 (31%)
`145 (69%)
`Mesalamine
`(n=1 64)
`
`Bleeding
`Yes
`
`No
`
`44 (27%)
`120 (73%)
`
`Placebo
`(n=96)
`
`44 (46%)
`52 (54%)
`
`P-Value
`
`<0.01
`
`(n=93)
`
`>
`
`39 (42%)
`
`54 (58%)
`
`
`
`
`
`
`
`‘
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Table 8: Mucosal A earance
`
`
`
`Study 3003
`
`Mucosal Appearance
`Normal
`Not Normal
`Study 3004
`
`Mesalamine
`(n=209)
`_
`
`Placebo
`(n=96)
`
`P-Value
`
`‘
`
`91 (44%)
`1 18 (56%)
`Mesalamine
`(n=l 64) ‘
`
`31 (32%)
`65 68%
`Placebo
`(n=93
`
`Mucosal Appearance
`Normal
`89 (54%)
`43 (46%)
`Not Normal
`75 46%
`50 (54%
`
`
`I also analyzed time-to-relapse for the PP populations. The Kaplan-Meier graphs for studies
`3003 and 3004 are presented in the Appendix. These results are exploratory but present another
`way of illustrating the lower relapse rates for the treated group. Notice the events extend beyond
`the double-blind period of six months.
`
`Given that Study MPUC3004 terminated before completing its originally planned enrollment, a
`sensitivity analysis was performed with the additional 43 subjects that the study would have
`enrolled had it not terminated. If we assume these subjects would have had a success rate equal
`to the observed placebo response of 68% (Table 5), then using the ITT population, the success
`rates would be 151/193 (78%) and 73/107 (67%) for Mesalamine and Placebo, respectively, p =
`0.06. This may suggest that the trial may have been stopped early to avoid a potential failure of
`its primary endpoint.
`‘
`
`13'
`
`

`

`4.
`
`FINDINGS IN SPECIAL/SUBGROUP POPULATIONS
`
`4.1 Gender, Race and Age and Other Special/Subgroup Populations
`Both studies were combined in order to facilitate the subgroup analyses. These analyses were,
`solely, for exploratory purposes. As noted in Table 9 below, a smaller effect size is indicated for
`the Russian subpopulation (p=0.042) which appears to be due to the higher placebo response
`rate. Efficacy is also weaker for the male population (p=0.039) again possibly due to the higher
`placebo response for males. In the subpopulation of age groups for 265 the results are not
`statistically significant (p=0.81) however,'the subgroup size is small for that age group. Efficacy
`results are consistent across baseline disease severity as well as time on remission.
`
`Table 9: Table Analysis of Primary Endpoint Variable (No-Relapse) by Subgroup (Both
`
`Studies Combined
`
`l30/172=75.5%
`166/201=82.6%
`
`l32/166=80%
`
`, 165/207=79%
`
`270/3 34=81%
`26/3 9=67%
`
`3
`‘ 274/344=80%
`
`69/101=68%
`
`50/88=57%
`
`103/166=62%
`16/23=70%
`
`'
`
`112/175=64%
`
`Age Group
`< 65
`2 65
`Race
`White
`
`i 21/28=75%
`
`7/14=50%
`
`Non-White
`Baseline Disease Severity Category* .
`Normal or None
`_>_]
`Baseline Remission Duration
`57 (61%)
`159 (78%)
`S 13 Weeks
`‘
`62 (65%)
`137 (81%)
`> 13 Weeks
`*Disease Severity: Sum of all DAI scores, ranging from 0 (normal) to 12 (maximum score).
`
`f 134/ 156=86%
`i 162/217=75%
`:
`
`47/72=65%
`72/1 l7=62%
`
`The table below shows the primary analysis cross-classified by both gender and country. These
`data suggest lack of efficacy for the Russia-males subgroup, again due to high placebo response.
`
`Table 10: Primary Endpoint Variable (No Relapse) by Country by Gender
`(Both Studies Combined)
`
`
`
`
`
`
`
`
`
`14
`
`
`
`
`
`62/82=76%
`
`68/90=76%%
`
`
`
`30/51=56%
`
`18/40=45%
`
` m7
`
`0/84=83%
`
`96/117=82%
`
`39/50=78%
`32/48=67%
`
`
`
`
`
`
`
`
`
`
`

`

`5.
`
`SUMMARY AND CONCLUSIONS
`
`5.1 Statistical Issues and Collective Evidence
`
`The primary efficacy endpoint comparisons showed a highly statistically significant difference
`between Mesalamine and placebo in study 3003 (p < .001). Study 3004, also, demonstrated
`statistically significant results (p = .029). When these analyses were repeated using the per-
`protocol population, the results were similar tothat of the intent-to-treat population.
`
`The sponsor’s imputation strategy assigned treatment failures to subjects who terminated the
`study early but only ifthe reason for drop-out was efficacy related or due to a UC-related
`adverse event. The reviewer performed the more conventional and conservative analysis of the
`ITT population assigning treatment failure to all subjects who terminated early. These results
`did not change the efficacy conclusions for study 3003; however, the efficacy results for study
`3004 were marginal (p = .046).
`
`The reviewer performed additional analyses of the primary endpoint, each analysis adjusting for
`a single factor: country, site, gender, age group (< 65 or 2 65), race (White/Non-White) and
`baseline severity category (0, 21). These results are consistent with the primary analysis for each
`study.
`
`T0 control for type I error, the sponsor planned a hierarchical testing strategy for the secondary
`' endpoints, although the order of testing was specified in a protocol amendment dated after
`completion of studies but prior to unblinding. For Study MPUC3003, only the first secondary
`endpoint (change from baseline in rectal bleeding score) shows statistically significant
`improvement from baseline. None of the secondary endpoints were statistically significant in
`Study MPUC3004.
`’
`
`The sample size for study 3004 was reduced in a late—stage protocol amendment. A total of 257
`patients instead of the planned 300 were analyzed. This could be interpreted as an unplanned or
`early st0pping ofthe study. Even though the decision appears to have been made prior to
`breaking the bli