CENTER FOR DRUG EVALUATION AND RESEARCH
`
`APPLICA TI0N NUMBER:
`
`7
`
`50-805
`
`STATISTICAL REVIEWg S!
`
`

`

` US. Department of Health and Human Services
`
`Food and Drug Administration
`Center for Drug Evaluation and Research
`Office of Phannacoepidemiology and Statistical Science
`Office of Biostatistics
`
`STATISTICAL REVIEW AND EVALUATION
`
`CLINICAL STUDIES
`
`NDA/Serial Number:
`
`50-805 / 000
`
`Drug Name:
`
`OraceaTM (doxycycline 'M .
`40 mg
`7
`
`capsules)
`
`Indication(s):
`
`Inflammatory Lesions of Rosacea
`
`Applicant:
`
`Date(s):
`
`Review Priority:
`
`Collagenex Pharmaceuticals, Inc.
`
`Received 8/01/2005, user fee (10 months) 6/01/2006
`Standard
`1
`
`Biometrics Division:
`
`Division 3, HFD—725
`
`Statistical Reviewer:
`
`Steve Thomson, HFD—725
`
`Concurring Reviewers:
`
`Team Leader: Mohamed Alosh, Ph. D., HFD—725
`
`Medical Division:
`
`Dermatology and Dental Products, HFD-54O
`
`Clinical Team:
`
`Clinical Reviewer: P. Brown, M.D.,HFD-540
`
`Team Leader: M. Luke, M.D., Ph.D.,HFD—540
`
`Project Manager:
`
`S. Jain, HFD-540
`
`Keywords:
`
`Analysis of covariance, Cochran—Mantel—Haenszel
`
`

`

`NDA 50-805 OraceaTM (doxycycline' a ) 40 mg
`
`CollaGenex Pharmaceuticals, Inc.
`
`Table of Contents
`
`1. EXECUTIVE SUMMARY ..................................................................................................................................... 5
`
`1.1
`1.2
`1.3
`
`CONCLUSIONS AND RECOMMENDATIONS ....................................................................................................... 5
`BRIEF OVERVIEW OF CLINICAL STUDIES ........................................................................................................ 6
`STATISTICAL ISSUES AND FINDINGS ............................................................................................................... 6
`
`2. INTRODUCTION ................................................................................................................
`
`
`OVERVIEW ...................................................................................................................................................... 7
`DATA SOURCES ............................................................................................................................................ 10
`
`2.1
`2.2
`
`3. STATISTICAL EVALUATION ......................................................................................................................... 11
`3.1
`EVALUATION OF EFFICACY .......................................................................................................................... 1 1
`3.2
`EVALUATION OF SAFETY .............................................................................................................................. 21
`
`4. FINDINGS IN SPECIAL/SUBGROUP POPULATIONS ............................................................................... 22
`
`4.1
`4.2
`
`GENDER, RACE AND AGE ............................................................................................................................. 23
`OTHER SPECIAL/SUBGROUP POPULATIONS: STRATIFICATION ON BASELINE LESION COUNT ...................... 26
`
`5. SUMMARY AND CONCLUSIONS ................................................................................................................... 27
`
`5.1
`5.2
`
`STATISTICAL ISSUES AND COLLECTIVE EVIDENCE ....................................................................... ,................ 27
`CONCLUSIONS AND RECOMMENDATIONS ..................................................................................................... 28
`
`APPENDICES: ..........................................................................................................................................................30
`
`APPENDIX 1. PER PROTOCOL ANALYSIS OF TOTAL LESION COUNT/CHANGE FROM BASELINE
`IN LESIONCOUNT.........................................................................................30
`
`APPENDIX 2. CHANGE FROM BASELINE IN INVESTIGATOR’S GLOBAL ASSESSMENT .................32
`
`APPENDIX 3. PER PROTOCOL ANALYSES OF THE INVESTIGATOR’S GLOBAL ASSESSMENT ....34
`
`APPENDIX 4. SENSITIVITY ANALYSIS TO CENTERS IN STUDY ROSE-301 ..........................................36
`
`APPENDIX 5. SENSITIVITY ANALYSIS TO CENTERS IN STUDY ROSE-302 ..........................................40
`
`APPENDIX 6. ASSOCIATION BETWEEN THE IGA AND THE INFLAMMATORY LESION COUNT ..43
`
`APPENDIX 7. SENSITIVITY ANALYSIS: EXTENDED INVESTIGATOR’S GLOBAL ASSESSMENT...44
`
`APPENDIX 8. REPORTED ADVERSE EVENTS BY NUMBER OF EVENTS AND NUMBER OF
`SUBJECTS WITH EVENT ...................................................................................................................................... 46
`
`APPENDIX 9. PRELIMINARY BAYESIAN ANALYSIS ................................................................................... 50
`
`

`

`NDA 50-805 OraceaTM (doxycych'ne __._\
`
`) 40 mg
`
`CollaGcnex Pharmaceuticals, Inc.
`
`LIST OF TABLES:
`
`TABLE 1. PATIENT DISPOSITION ............................................................................................. 14
`
`TABLE 2. SUBJECT DEMOGRAPHICS .....................................................................................I ..... 14
`
`TABLE 3. ROSE-301 (ITT-LOCF) MEAN CHANGE FROM BASELINE IN LESION COUNT ............................... 15
`
`TABLE 4. ROSE-302 (ITT-LOCF) MEAN CHANGE FROM BASELINE IN LESION COUNT .............................. 16
`
`TABLE 5. ROSE-301 (ITT-LOCF) INVESTIGATOR’S GLOBAL ASSESSMENT ............................................ 17
`
`TABLE 6. ROSE-302 (ITT-LOCF) INVESTIGATOR’S GLOBAL ASSESSMENT ............................................ 18
`
`TABLE 7. ROSE-301 (ITT-LOCF) MEAN CHANGE FROM BASELINE IN ERYTHEMA SCORE.......................... 19
`
`TABLE 8. ROSE-302 (ITT—LOCF) MEAN CHANGE FROM BASELINE IN ERYTHEMA SCORE .........................20
`
`TABLE 9. SPONSOR RESULTS ON INFLAMMATORY LESIONS (ITT POPULATION) .......................................20
`
`V TABLE 10. SPONSOR REPORTED SUCCESS ON INVESTIGATOR’S GLOBAL ASSESSMENT (ITT) .......................21
`
`TABLE 1 1. SPONSOR RESULTS ON ERYTHEMA ASSESSMENT SCORE (ITT POPULATION). . .. ......................... 21
`
`TABLE 12. CHANGE IN TOTAL INFLAMMATORY LESIONS SCORE BY GENDER BY WEEK...................._ ......... 23
`
`TABLE 13. WEEK 16 INVESTIGATOR GLOBAL ASSESSMENT BY GENDER ..............................................23
`
`TABLE 14. CHANGE IN TOTAL INFLAMMATORY LESIONS SCORE BY RACE BY WEEK............................... .24
`
`TABLE 15. WEEK 16 INVESTIGATOR GLOBAL ASSESSMENT BY RACE ................................................. 24
`
`TABLE 16. CHANGE IN TOTAL INFLAMMATORY LESIONS SCORE BY AGE GROUP BY WEEK ........................ 25
`
`TABLE 17. WEEK 16 INVESTIGATOR GLOBAL ASSESSMENT BY AGE GROUP .......................................... 25
`
`TABLE 18. CHANGE IN TOTAL INFLAMMATORY LESIONS SCORE BY BASELINE SCORE BY WEEK ...................26
`
`TABLE 19. WEEK 16 INVESTIGATOR’S GLOBAL ASSESSMENT BY BASELINE SCORE .................................. 26
`
`TABLE A.1.1 ROSE-301 (PER PROTOCOL) MEAN CHANGE FROM BASELINE 1N LESION COUNT ..................... 30
`
`TABLE A. 1 .2_ROSE—302 (PER PROTOCOL) MEAN CHANGE FROM BASELINE IN LESION COUNT .......................... 31
`
`TABLE A.2.1 ROSE-301: CHANGE FROM BASELINE IN IGA ............................................................. .32
`
`TABLE A.2.2 ROSE-302: CHANGE FROM BASELINE IN IGA ..............................................................33
`
`TABLE A3. 1. ROSE-301 (PER PROTOCOL) INVESTIGATOR’S GLOBAL ASSESSMENT ................................. 34
`
`TABLE A.3.2. ROSE—3 02 (PER PROTOCOL) INVESTIGATOR’S GLOBAL ASSESSMENT ................................. 35
`
`

`

`NDA 50-805 OraceaTM (doxycycline
`
`(_..— ) 40 mg
`
`CollaGenex Pharmaceuticals, Inc.
`
`TABLE A.4.1 NUMBER OF SUBJECTS PER CENTER (ITT-LOCF) IN ROSE-301 ......................................... 36
`
`TABLE A.4.2 MEAN CHANGE IN LESION COUNTS PER CENTER (ITT-LOCF) IN ROSE-301 .......................... 36
`
`- TABLE A.4.3 EFFECT OF DELETING CENTERS ON LESION COUNTS (ITT-LOCF) IN ROSE-301 ......................36
`
`TABLE A.4.4 DISTRIBUTION OF IGA AND THE EFFECT OF DELETING CENTERS IN ROSE-301 ....................... 37
`
`TABLE A.4.5 DISTRIBUTION OF CHANGE FROM BASELINE IN IGA AND THE EFFECT OF DELETING CENTERS IN ROSE-
`301 ................................................................................................................................. 39
`
`TABLE A.5.1 NUMBER OF SUBJECTS PER CENTER (ITT—LOCF) IN ROSE-302 ......................................... 40
`
`TABLE A.5:2 MEAN CHANGE IN LESION COUNTS PER CENTER (ITT—LOCF) IN ROSE-302............................40
`
`TABLE A.5.3 EFFECT OF DELETING CENTERS ON LESION COUNTS (ITT-LOCF) IN ROSE-302 .......................40
`
`TABLE A.5.4 DISTRIBUTION OF IGA AND THE EFFECT OF DELETING CENTERS IN ROSE-302 ........................41
`
`TABLE A.5.5 DISTRIBUTION OF CHANGE FROM BASELINE IN IGA AND THE EFFECT OF DELETING CENTERS IN ROSE-
`302 ................................................................................................................................. 42
`
`TABLE A.6.1 ROSE-301 ASSOCIATION BETWEEN THE WEEK 16 IGA AND TOTAL LESION COUNT ....................43
`
`TABLE A.6.2 ROSE-302 ASSOCIATION BETWEEN THE WEEK 16 IGA AND TOTAL LESION COUNT .................... 43
`
`TABLE A.7.1. ROSE-301 EXTENDED INVESTIGATOR’S GLOBAL ASSESSMENT. . .
`
`. .
`
`.................................44
`
`TABLE A.7.2. ROSE-302 EXTENDED INVESTIGATOR’S GLOBAL ASSESSMENT..........................................45
`
`TABLE A.8.1 ADVERSE EVENTS: NUMBER OF EVENTS AND NUMBER OF SUBJECTS .................................._...46
`
`TABLE A.9.1 ROSE—301 SUMMARIES OF POSTERIOR DISTRIBUTIONS ......................................... '.......... 50
`
`TABLE A.9.2 ROSE-302 SUMMARIES OF POSTERIOR DISTRIBUTIONS .................................................... 51
`
`

`

`NDA 50-805 OraceaTM (doxycycline \ , 40 mg
`
`'
`
`CollaGenex Pharmaceuticals, Inc.
`
`1. EXECUTIVE SUMMARY
`
`OraceaTM (doxycycline \ capsules) 40 mg is intended for
`oral administration to be taken once daily to ' / inflammatory lesions in patients with
`rosacea. The Sponsor submitted the results from two Phase 3 studies, COL—101-ROSE—301 and
`COL-102-ROSE—302 (referred to as either ROSE—301 and ROSE-302, or Study 301 and 302,
`respectively). This report summarizes the analyses of these studies.
`
`1.1 Conclusions and Recommendations
`
`For both Phase 3 studies the primary efficacy endpoint specified in the protocols was the
`change from baseline in inflammatory lesion count.
`In Study ROSE-301, at baseline the mean
`number of lesionsin the Oracea group was 19. 5 versus 20.31n the Placebo group. In the intent-
`to—treat (ITT) population, using last observation carried forward (LOCF) imputation for
`dropouts, the Week 16 mean changes from this baseline count were-11.8 and —.5 9 respectively
`Using a simple ANOVA model, the difference1n this change from baseline was statistically
`significant (p < 0.0020).
`In Study ROSE—302, at baseline the mean numbers of lesions were
`20. 5 and 212 in the Oracea and Placebo groups, respectively. So in both studies, while the
`difference was not statistically significant, the baseline lesion count was higher (1.e. ,worse)1n
`the placebo group than the corresponding Oracea group. In ROSE-302, the corresponding Week
`16 mean changes from this baseline count were——.9 5 and —.4 3, respectively Again, as in ROSE-
`301, the difference in these changes from baseline was statistically significant (p < 0.0001). A
`preliminary Bayesian analysis using growth curve models confirmed these results By week 12,
`the posterior probability at least 0.98 that a patient using Oracea Would be expected to have at
`least three lesions less than when using Placebo (Please see Appendix 9).
`
`In response to a request by the Division for a static Investigator Global Assessment
`(IGA), measuring global rosacea (except possibly erythema), the Sponsor provided an endpoint
`thatIS basically a grouped data version of the inflammatory lesion count (Please see section 2.1. 2
`and Appendix 6 for more on this.) At Week 161n the intent—to-treat (ITT) population1n Study
`ROSE—301, according to the Sponsor 5 IGA,11 of the 127 Oracea patients versus 10 of the 124
`Placebo patients were clear, i e. ,had no inflammatory lesions For statistical analysis, the ”
`Division recommended dichotomizing this endpoint so that a “success” was defined as an IGA of
`“Clear” or “Near Clear.” At Week 16, in ROSE—301, 16.5% of the Oracea patients and 10.4% of
`the Placebo patients were scored as successes on this endpoint. In Study ROSE-302, at Week
`16, only two of the 142 Oracea patients versus none of the 144 Placebo patients were scored as
`“Clear,” while 8.5% of the Oracea patients versus 3.4% of the Placebo patients were scored as
`“Clear” or “Near Clear.” These treatment differences in success rates were statistically
`significant (p S 0.0361 and p S 0.012, in Studies ROSE-301 and ROSE—302, respectively).
`attempt to provide a static overall IGA the Medical team defined a post hoc extended IGA,
`incorporating erythema. (please see Appendix 7). There were no statistically significant
`treatment differences on this endpoint
`
`In an
`
`

`

`NDA 50-805 OraceaTM (doxycycline
`
`,
`
`)40 mg
`
`CollaGenex Pharmaceuticals, Inc.
`
`1.2 Brief Overview of Clinical Studies
`
`Studies ROSE—301 and ROSE-302 were randomized, double—blind, placebo—controlled,
`parallel group, multicenter, 16 week Phase 3 trials conducted in the United States with a total of
`537 rosacea patients in both studies, 269 of whom were treated with Oracea. Study ROSE—302
`included a further four week extension without treatment. Patients were randomized 1:1 to
`
`Oracea and Placebo. The study protocols defined the total inflammatory lesion counts as the sum
`of papule, pustule, and nodule counts. The primary efficacy endpoint was the Week 16 change
`from baseline in this inflammatory lesion count. The Division also requested that an Investigator
`Global Assessment (IGA) of the overall rosacea status be defined as a co-primary endpoint. The
`Sponsor argues that since the proposed indication is “to \ inflammatory lesions in patients
`with rosacea” only the change from baseline in lesion count should be used as a primary
`endpoint. The analyses presented here follow the original Division recommendation.
`
`1.3 Statistical Issues and Findings
`
`Statistical Issues
`
`Perhaps the most important issue with this submission is whether or not the IGA is
`1.
`defined appropriately, and is suitable as a primary endpoint. As discussed in section 2.1.2 for a
`rosacea indication, the division requested a global assessment of rosacea. The Sponsor provided
`an endpoint that is a grouped data version of the lesion count in Study 301 and largely a grouped
`data version in Study 302. Note the Sponsor argues that since the proposed indication is “to
`—_—< inflammatory lesions in patients with rosacea” only the change from baseline in lesion
`count is needed as a primary endpoint. Both endpoints were analyzed in this review.
`
`The protocols specify that the changes from baseline in lesion counts are to be analyzed
`2.
`with an analysis of variance (ANOVA) model with factors for treatment and center. The
`protocol also specified that if the residuals are not normal a van Elteren test is to be used to
`compare median scores of the treatment group. This reviewer’s opinion is that in all cases in
`these studies the data do not seem to be sufficiently skewed to invalidate the assumption of
`approximate normality in the distribution of cell means. Thus ANOVA would be appropriate.
`However since this was specified in the protocol, results from both statistical tests are reported
`here, and are always essentially equivalent.
`
`The IGA was measured on a 0—4 scale, but the guidelines indicate the corresponding
`3.
`associated range of inflammatory lesions are 0, 1—2, 3-10, 11-19, or 20+, respectively. For
`analysis the Division recommended dichotomizing the IGA so that treatment “success” was
`defined as a score of 0 or 1, otherwise it was a “failure”. This endpoint was used in the analyses
`in this report. The Sponsor’s prOtocol indicates that the second analysis of this endpoint should
`be based on the change from baseline. This change from baseline is summarized in Appendix 2;
`however, note that the difference in IGA scores between 0 and l (i.e., between 0 and l—2 lesions)
`is not particularly commensurable with the nominally equal difference between 3 and 4 (i.e.,
`
`

`

`NDA 50-805 OraceaTM (doxycycline - M i 40 mg
`
`CollaGenex Pharmaceuticals, Inc.
`
`between 11-19 and 20+ lesions). This would suggest that the change from baseline in this
`endpoint is not a particularly usefiJl measure.
`
`Several centers recruited only a small number of patients into the study. Pooling of
`4.
`subjects for the analysis was specified in amendment to the protocol issued on April 26, 2005,
`after completion of both studies. This is clearly a post hoc adjustment. However, this pooling
`was deemed to be acceptable, and for convenience was followed in the Agency analysis.
`
`.
`
`Statistical Findings
`
`In Study ROSE—301, at baseline the mean number of lesions in the Oracea group was
`19.5 versus 20.3 in the Placebo group. In Study ROSE-302, at baseline the mean numbers of
`lesions were 20.5 and 21.2 in the Oracea and Placebo groups, respectively. So in both studies,
`while the difference was not statistically significant, the baseline score was higher in the Placebo
`group. In the ITT—LOCF ROSE—301 population, the Week 16 mean changes from this baseline
`count were —11.8 and -5.9 for Oracea and Placebo, respectively. Using a simple ANOVA model,
`the difference in this change from baseline was statistically significant (p S 0.0002). In ROSE-
`302, the corresponding Week 16 mean changes from this baseline count were -9.5 and -4.3,
`respectively. Again, as in Study 301, the difference in these changes from baseline was
`statistically significant (p < 0.0001).
`
`At Week 16 in the ITT population in Study ROSE-301, according to the Sponsor’s IGA,
`11 0f the 127 Oracea patients versus 10 of the 124 Placebo patients were clear, i.e., had no
`inflammatory lesions. For analysis, the Division recommended dichotomizing this endpoint so
`that a “success” was defined as an IGA of “Clear” or “Near Clear.” At Week 16, in Study 301,
`16.5% of the Oracea patients and 10.4% of the Placebo patients were scored as successes on this
`endpoint. In Study ROSE-302, at Week 16, only two of the 142 Oracea patients versus none of
`the 144 Placebo patients were scored as “Clear,” While 8.5% of the Oracea patients vers1'1s 3.4%
`of the Placebo patients were scored as “Clear” or “Near Clear.” These treatment differences in
`
`success rates were also statistically significant (p S 0.0361 and p S 0.012, in Studies ROSE-301
`and ROSE—302, respectively).
`
`Results were generally consistent in the Per Protocol population and seemed to be
`generally consistent for each gender, and overall, across age groups. Few patients were non—
`Caucasian, butamong these few patients, there was no particular evidence that treatment efficacy
`was greater than placebo.
`
`2. INTRODUCTION
`
`2.1 Overview
`
`According to the Sponsor: “The clinical development program for OraceaTM included two
`pilot pharmacokinetic studies, a multiple-dose steady state bioequivalence study comparing
`OraceaTM with Periostat®, a food—effect study, and two Phase 3 studies, Oracea—ROSE-301 and
`7
`
`

`

`NDA 50-805 OraceaTM (doxycycline, "“8 40 mg
`
`CollaGenex Pharmaceuticals, Inc.
`
`Oracea-ROSE-302 (referred to as 301 and 302, or ROSE-301 and ROSE-302, respectively). All
`of these studies used the same formulation of the drug product proposed for marketing. The
`randomized, double-blind, placebo-controlled, parallel group, multicenter, Phase 3 trials included
`537 patients with rosacea, 269 of whom were treated with OraceaTM for up to 16 weeks. All of
`these studies were conducted in the United States.” (page 78, volume 1.1, module 2)
`
`' 2.1.1 Design
`
`'Both Studies ROSE-301 and ROSE—302 are described as multicenter, randomized,
`double-blind placebo controlled, parallel group, Phase 3 studies, each conducted at 14
`investigational centers. Patients were randomly assigned 1:1 to OraceaTM (doxycycline
`
`\ capsules) to be taken for 16 weeks. Patients were evaluated at
`' baseline, and at Weeks 3, 6, 12, and 16. Study 302 included an extra evaluation at Week 20.
`Study 301 was initiated on 22 June 2004 and the last patient completed on 1 April 2005. Study
`302 was initiated on 24 June 2004 and completed on 4 April 2005. Please see Section 2.1.2,
`below, for details on the regulatory history. Summaries of patient disposition and demographics
`are given in Section 3.1.2.
`
`2.1.2 Regulatory History:
`
`Pre-IND/End of Phase 2 Meeting (January 28, 2002), FDA minutes (sent to the Sponsor
`1.
`on February 13, 2002):
`
`The Sponsor initially requested a claim for both R and rosacea, requesting one
`study of each condition. However, the FDA stated that since these were considered to be
`separate diseases, “The Sponsor should conduct two, adequate, placebo—controlled trials for each
`indication.” (page 3 of minutes)
`
`Further, for an indication of rosacea:
`“i. The Agency supports the following two primary efficacy endpoints: the Investigator’s
`(Clinician’s) Global Assessment and lesion counts. The Investigator’s Global Assessment
`should be a static assessment at efficacy endpoint and not a change from baseline. The
`Investigator’s Global Assessment should be dichotomized a priori to success and failure. As
`presented1n the briefing package the Agency would support success as a score of‘ 0’ on the
`assessment scale” (page 4 of minutes)
`
`“If the Sponsor can be more precise in its description of the difference between score 1 and 2,
`then the Agency might consider adding category 1 to success.” (page 4 of minutes)
`-
`
`“ii. There should be a statistically significant reduction in inflammatory lesions at endpoint. For
`approval, success must be demonstrated in both the Investigator’s Global Assessment and in
`lesion counts.” (page 4 of minutes)
`
`

`

`NDA 50-805 OraceaTM (doxycycline
`
`
`
`40 mg
`
`CollaGenex Pharmaceuticals, Inc.
`
`Pre-IND/End of Phase 2 Meeting (May 3, 2004), FDA minutes (sent to the Sponsor on
`2.
`May 27, 2004):
`
`The Sponsor indicated that the requested indication was the treatment of papules and pustules of
`,-——-_-~ rosacea, but not erythema. However, the Division commented that the
`Sponsor did not conduct adequate dose ranging. Also, the Division stated that “The Agency
`recommends that the Clinician’s Global Severity Score be modified to include static clinical
`descriptors and categories (e.g., Clear, Almost Clear, Mild, Moderate, and Severe). The
`Clinician’s Global Severity Score appears to be similar to an Investigator’s Global Assessment
`(IGA) scale; however, as an IGA the Agency recommends use of clinical descriptors (e.g.,
`papules, nodules, slight pinkness, fiery redness, telangiectasia, etc.) The Sponsor’s Clinician’s
`Global Severity Score includes an area specific “score” which is not a clinical global
`assessment.” (page 5 of minutes)
`
`The Agency agreed that erythema could be removed from the Clinician’s Global Severity
`Scale and be evaluated as a secondary endpoint. Note that the Clinician’s Global Severity Scale
`was renamed to the Investigator’s Global Assessment in this submission, but is largely or
`essentially only a grouped data version of the lesion count.
`
`Further, the Agency reminded the Sponsor that “The endpoints are not as recommended
`at the January 28, 2002, Pre-IND/End of Phase 2 Meeting to support the rosacea indication.
`i. The Agency recommends the following two primary efficacy endpoints for demonstrating
`efficacy in treatment of rosacea: 1) inflammatory lesion counts (papules, pustules, and nodules)
`and 2) the investigator's static global assessment (IGA). Clinical signs (erythema and
`telangiectasia) should be incorporated into the static global assessment.
`ii. As noted above, the Agency recommends that the IGA be a static scoring system.‘ The IGA
`should be dichotomized a priori to success and failure.
`iii. For approval, success must be demonstrated in both the IGA and in lesion counts. There
`should be a statistically significant reduction in inflammatory lesions at study endpoint.
`iv. The Sponsor proposes a Clinician’s Erythema Score .
`.
`. obtained at endpoint as a sum
`obtained from evaluation of five facial areas (scale of 0 to 4). The Sponsor is reminded that if a
`reduction of erythema is sought as part of the indication, then this parameter should be
`incorporated into the IGA.” (page 6 of minutes)
`
`Further, in the Biostatistics comments:
`“The protocol includes a large number of secondary endpoints. The Sponsor should consider a
`limited number of clinically relevant endpoints or an adjustment for multiplicity may be needed.
`During the meeting the Sponsor said that it could classify clinically relevant secondary endpoints
`into two groups: a small number that might be considered for labeling, and those with only
`exploratory interest.” (page 7 of minutes)
`
`The Protocol review dated September 27, 2004, reiterated these comments. It was noted
`3.
`that “instead of a secondary endpoint the IGA should be defined as a co—primary endpoint.”
`Further, telangiectasia should be included in the IGA,
`Tm“,
`
`9
`
`

`

`NDA 50—805 OraceaTM (doxycycline V——-
`
`40 mg
`
`CollaGenex Pharmaceuticals, Inc.
`
`Pre—NDA Meeting (March 30, 2005), FDA minutes (sent to the Sponsor on May 27,
`
`4.
`2005):
`
`The Division again stated that the “analysis of the dichotomized IGA as a secondary
`variable is not acceptable. The Agency stands by the recommendation for use of co-primary
`efficacy endpoints for rosacea provided to the sponsor at the January 28 2002, Pre—IND/End of
`Phase 2 meeting, May 3, 2004, End of Phase 2 Meeting, and protocol comments of September
`27, 2004. As it is too late to modify the prespecified analysis plan, the Agency recommends the
`following analyses be submitted.
`a. Submit data analysis as pre-specified in your statistical analysis plan in your protocol.
`b. Submit data analysis as was recommended by the Agency.” (pages 7-8 of minutes)
`
`“Two primary efficacy endpoints are needed for demonstrating efficacy in treatment of rosacea:
`a): inflammatory lesion counts .
`.
`; and b) the investigator’s static global assessment (IGA).
`.
`. For approval, success must be demonstrated in both the Investigator’s Global Assessment
`and in lesion counts. Subjects enrolled with an IGA in the win category (i.e. clear or almost
`clear) should not be included in the analysis. It was discussed that erythema and telangiectasia
`are not included in the Investigator’s global assessment scale, will be addressed as secondary
`variables and should not get worse.” (page 8 of minutes) Note that no such subjects with an
`IGA in the success category were actually enrolled at baseline, and that the Division thus
`confirmed that the clear and almost clear categories in the IGA are to be used to define
`“success”.
`
`The Sponsor states (in their response to the FDA 74 day letter, received 2 November
`5.
`2005) that “Per the instruction of Dr. Jonathan Wilkin, Division Director, HFD—540 during the
`Pre-NDA meeting, the Sponsor was to maintain the IGA‘ as a secondary endpoint and file the
`results as requested by the Agency.” This particular claim does not seem to be confirmed by the
`FDA minutes of that meeting,
`
`,__._ H
`The original discussion for this submission seems to be addressed for a
`.
`5; However, the Sponsor argues that since the proposed indication1s more limited, i. e. ,“to
`{/‘rinflammatory lesions1n patients with rosacea’’only the change from baselinein lesion
`count should be used as a primary endpoint. This analysis follows the original Division
`recommendation and uses both primary endpoints.
`
`2.2 Data Sources
`
`Data for the pivotal study was downloaded from the FDA Electronic Data Room as SAS
`transport files, located in the following link:
`
`
`\\CDSESUB1\N50805\N 000\2005-07—29
`
`10
`
`

`

`NDA 50-805 OraceaTM (doxycycline‘v '/ 40 mg
`
`CollaGenex Pharmaceuticals, Inc.
`
`3. STATISTICAL EVALUATION
`
`3.1
`
`Evaluation of Efficacy
`
`Efficacy results are based on the data from two similar Phase 3 studies, labeled COL—101-
`ROSE-301 and COL-101-ROSE-302 (i.e., Studies ROSE-301 and ROSE—302, or 301 and 302),
`respectively, each study titled:
`
`A Multicenter, Randomized, Double-Blind, Placebo-Controlled, Parallel-Group Clinical Trial to
`Determine the Effects of 40 mg Doxycycline Monohydrate Modified Release Capsules (COL—
`101) Administered Once Daily Versus a Placebo Control Administered Once Daily for the
`Treatment of Rosacea.
`
`The Sponsor reports that ROSE-301 was initiated on June 22, 2004, and completed April 1,
`2005, while ROSE-302 was initiated on June 24, 2004, and completed April 4, 2005. The
`Sponsor reports that final protocols for both studies were issued on May 7, 2004.
`
`3.1.1 Study Design and Endpoints
`
`Two very similar Phase 3 studies, ROSE-301 and ROSE-302, or for brevity labeled as
`Studies 301 and 302, respectively, were conducted. The only difference between the two studies
`was that ROSE-302 included a 4—week extension period without treatment. The Sponsor
`describes these as: “Both studies were outpatient, multicenter, randomized, double—blind,
`placebo-controlled, parallel group trials to evaluate the safety and efficacy of OraceaTM for
`reducing total inflammatory lesions compared with placebo. Patients were to take one capsule of
`study medication once daily every morning for 16 weeks. Study visits were at Baseline and
`Weeks 3, 6, 12, and 16, and in Study 302 also at Week 20 (patients stopped treatment at Week
`16).”
`(page 78, volume 1.1, module 2) Patients were randomized 1:1 to Oracea or placebo (i.e.,
`vehicle).
`
`Primary Efficacy Endpoints
`
`The primary efficacy endpoint identified in the protocols was the Week 16 change from
`baseline in total inflammatory lesion count. Total inflammatory lesion count was defined as the
`sum of papule, pustule, and nodule counts. As discussed in Section 2.1.2, the Division
`recommended a static Investigator Global Assessment as a co-primary endpoint. The Sponsor
`argues that since the proposed indication is “to e” inflammatory lesions in patients with
`rosacea” only the change from baseline in lesion count should be used as a primary endpoint.
`This analysis will follow both the original Division recommendation and the protocol definition.
`
`The Investigator’s Global Assessment was measured at Weeks 3, 6, 12, and 16, and is I
`defined as follows. Note that for entry to the studies patients had to score at least a “2” (i.e., a
`score of “Mild”) on the IGA. Most patients entered with a score of 3 (i.e., “Moderate”).
`
`ll
`
`

`

`NDA 50-805 OraceaTM (doxycycline _ ,/ :) 40 mg
`
`CollaGenex Pharmaceuticals, Inc.
`
`
`
`Investi_’ator s Global Assessment (IGA):
`
`
`Definition
`guideline
`—I
`
`
`No signs or symptoms present
`’
`Skin clear of inflammatory lesions
`7
`
`One or two papules
`1 or 2 small, non—inflammatory lesions
`
`
`Some papules/pustules
`3 to 10 papules/pustules
`'
` Moderate
`Moderate number of papules/pustules
`11 to 19 papules/pustules
`
`Numerous papules/pustules; nodules
`I 220 papules/pustules and nodules
`
`
`
`
`For the analysis, following the Division recommendation, “success” on this endpoint is
`defined as a grade of “Near Clear” or “Clear”, i.e., a score of 0 or 1. However, this endpoint,
`unlike the IGA recommended by the Division, is primarily a grouped data version of the
`inflammatory lesion count. With a single exception in Study ROSE-301, at each visit the lesion
`counts fall within the ranges assigned by the IGA.
`In Study ROSE—302, the matching between
`ranges of lesion counts and levels of the IGA is somewhat less consistent, but generally the _
`lesion counts also fall within the ranges assigned by the IGA. Following the Division
`recommendation, the Week 16 score on this assessment is considered as a primary endpoint.
`
`The Sponsor’s analysis is based on the change from baseline in this IGA. Note that the
`computed differences in the IGA scores do not correspond to equal counts in lesions. For
`example, in the “natural” metric a one unit difference between IGA scores of 0 and 1 correspond
`to 1 or 2 lesions,_while say a one unit difference between IGA scores of 2 and 3 correspond to
`between 1 to 16 lesions. This suggests that an analysis of the actual IGA scores would be more
`interpretable than the analysis based on change from baseline. For the FDA analysis the actual
`scores, dichotomized so that “success” is an IGA of O or 1, is used in the primary analysis,
`however the protocol specified analysis using change from baseline is provided in Appendix 2.
`
`Secondary Efficacy Endpoints
`An erythema score was defined for each of the forehead, chin, nose, right check, and left
`check, each facial region assessed on