CENTER FOR DRUG EVALUATION AND
`
`RESEARCH
`
`APPLICA TION NUMBER:
`
`22-341
`
`STATISTICAL REVIEWg S2
`
`

`

`STATISTICAL REVIEW AND EVALUATION
`
`CLINICAL STUDIES
`
`NDA/Serial Number:
`
`022341/0
`
`Drug Name:
`
`VictozaTM (liraglutide) injection
`
`Indication(s):
`
`Treatment of type 2 diabetes mellitus
`
`Applicant:
`
`Date(s):_
`
`Novo Nordisk Inc
`
`July 8, 2009
`Submission date:
`Review date: September 8, 2009
`
`Review Priority:
`
`Standard
`
`Biometrics Division:
`
`Division of Biometrics 2
`
`Statistical Reviewer:
`
`Janice Derr, Ph.D.
`
`Statistics Team Leader:
`
`J. Todd Sahlroot, Ph.D.
`
`Medical Reviewer
`
`Karen M. Mahoney, MD.
`
`Medical Team Leader
`
`Hylton Joffe, M.D.
`
`Project Manager
`
`John Bishai, Ph.D.
`
`This memorandum is the statistical review of the protocol of Study 3748, which is proposed to
`evaluate the long-term effect of liraglutide on cardiovascular and other clinically important
`outcomes. A brief description of the statistical aspects of the design and proposed analysis is
`included at the end of this memo.
`.
`
`Statistical review comments, to be transmitted to the sponsor:
`
`1. We agree with the calculations of the number of subjects needed in the study, subject to
`clinical input on the appropriateness of the assumption of a 1.8% event rate per year in this
`clinical population. Our understanding is that the study is designed to accumulate a total of
`approximately 611 adjudicated primary outcome events across the two study arms. Please
`confirm or clarify this total.
`
`2. We have the following requests concerning the proposed interim assessment of efficacy:
`
`(a) The protocol should specify the number of events associated with the two proposed
`interim assessments. We assume that 50% of the expected number of events is
`approximately 306 and 75% is approximately 458. Please confirm or clarify this
`assumption.
`
`(b) The protocol should specify which efficacy outcome variable(s) will be assessed for
`superiority, using the modified Haybittle—Peto stopping boundary.
`If more than one
`outcome variable will be assessed, the protocol should provide more information about
`the protection of Type I error for the primary cardiovascular outcome variable.
`
`

`

`(c) The protocol should specify how the interim analysis of efficacy and futility will be
`conducted, in order to maintain the appropriate study blind.
`
`3. The statistical methods for the analysis of primary and supportive outcome data that are
`generally described in this protocol are acceptable.
`In addition, we request that you submit the
`more detailed statistical analysis plan with sufficient lead time prior to your analysis of data so
`that we may review the plan and send you our review comments.
`
`4. Study 3748 presents an opportunity to gain further information concerning the comparison
`between liraglutide and placebo in longitudinal changes in serum calcitonin in this study
`population. We recommend that
`the study protocol
`include a detailed analysis plan for
`evaluating this relationship. This analysis plan should include a pre—specified statistical analysis
`model, along with additional supportive analyses and descriptive summaries.
`
`

`

`design (not to be transmitted to the sponsor):
`Summary of the study
`5; Evaluation of
`EX221-3748, “Liraglutide Effect and Action in Qiabete
`mized, double—
`Title of Study:
`come Results; A five—year, multi—centre, international, rando
`scular events”
`cardiovascular out
`d trial
`to determine liraglutide effects on cardiova
`blind, placebo—controlle
`(LEADER).
`
`Purpose: The primary purpose ofthis study is to determine the long—term effect ofliraglutide on
`cardiovascular and other clinically important outcomes.
`r 2 oral anti—diabetes drugs (OAD)
`abetes treated with 0, 1 0
`Trial design: Subjects with type 2 di d of a minimum of tw
`0 weeks, be randomized (1:1) to
`will, after a single-blind run-in perio
`dd—on to their standard of care
`receive liraglutide 1.8 mg once daily or equivalent placebo as an a
`atients among type 2 diabetic subjects
`treatment. The study will enroll approximately 9000 p ’tment period is planned for 18 months,
`who are at high risk for cardiovascular events. The recrui
`and intended maximum trial duration will be 60 months. The minimum duration of observation
`after randomization will be 42 months.
`
`0 be randomized in the study was
`
`Number of subjects in the study: The number of subjects t
`estimated based on a time to first outcome using a log rank test on an intention—to—treat analysis
`and the following assumptions: a) a conservative range of primary outcome event rate of 1.8%
`per year; b) a 1 sided alpha of 0.025; c) uniform enrolment over 1.5 years with a maximum
`f 5 years (including the accrual period) ; d) a non—inferiority margin versus placebo
`of 1.3 for the upper limit ofthe 2-sided 95% confidence interval; 6) a non-adherence rate of 10%;
`by the second year in trial and uniform thereafter; and f) 90% power to reject the null hypothesis
`that the hazard ratio is > 1.3.
`ptions, 8900 subjects need to be rando
`Under the above assum
`liraglutide with high power.
`cardiovascular effects of
`(approximately) using the
`Statistical review comment:
`I was able to recreate this calculation
`otal of 611 adjudicated
`statistical software EastTM5.2. The study is designed to accumulate a t
`primary 01
`rms combined.
`ttcome events in both study a
`
`mized to clearly evaluate the
`
`ence of either cardiovascular (CV) death,
`me is the first occurr
`Secondary outcomes include an
`Outcomes: The primary outco
`(M1) or nonfatal stroke.
`lar outcome, and all—cause mortality.
`nonfatal myocardial
`infarction
`mposite microvascu
`edical events of special
`expanded composite of CV events, a co
`s adverse events and other m
`Among the other endpoints are seriou
`ents leading to treatment
`d disease and adverse ev
`interest such as pancreatitis, neoplasms, thyroi
`6 also included as other
`discontinuation. HbAlc and laboratory endpoints such as calcitonin ar
`
`endpoints.
`
`ed for screening, the start of run—in, and
`plann
`Pre—treatment clinical visits are
`treatment period are planned for week 2, then
`Assessment:
`Visits during the
`and 60. Assessments include composite
`randomization (baseline).
`24, 30, 36, 42, 48, 54
`rtality, individual components of the
`months 1, 3, 6, 12, 18,
`ar outcomes, all—cause mo
`cardiovascular and microvascul
`rovascular outcomes, weight and waist—to—hip ratio, sustained
`'
`cardiovascular and mic
`function, serious adverse events and
`ypoglycaemia, cognitive
`3
`
`

`

`erest (pancreatitis, neoplasm, thyroid—related events,
`medical adverse events of special int
`ressure and heart rate, as well as
`other
`0 treatment discontinuation), blood p
`ted safety parameters (including
`adverse events leading t
`blood lipids, HbAlc and selec
`selected laboratory parameters:
`calcitonin, amylase and lipase).
`
`rnal DMC will be constituted for
`ittee (DMC): An independent exte
`nts, and to provide advice
`Data Monitoring Comm
`fety surveillance at pre—defmed time poi
`dify or terminate the trial
`the trial to perform ongoing sa
`t of the trial as whether to continue, mo
`-blinded. Clear evidence
`to the sponsor during the conduc
`vant safety information un
`as necessary. The DMC will evaluate all rele
`ther variables identified
`of net harm with respect to total mortality, cancer, hospitalizations or 0
`ver time and
`her studies, that is consistent o
`by the DMC based on emerging data from this or ot
`across subgroups would justify a recommendation to stop the trial early.
`ommittee (EAC): An external EAC will be constituted for the trial to
`dication, standardization and assessment of pertinent events in an
`Event Adjudication C
`manner,
`including cardiovascular death, acute coronary syndrome,
`perform ongoing adju
`y requiring acute hospitalization, pancreatitis, and all neoplasms.
`independent and blinded
`stroke, cardiac insufficienc
`es their intention to maximize
`ysis: The sponsor describ 5 who prematurely discontinue
`Evaluability of subjects for anal d to follow up with subject
`adherence to the study protocol, an
`their assigned treatment.
`more detailed statistical analysis
`Statistical considerations: The sponsor plans to prepare a
`d before the database is released. No analyses of unmasked or
`re the database is closed or released, except for confidential
`plan, which will be finalize
`between-group data is planned befo
`nt Data Monitoring Committee.
`analyses performed to support the deliberations of the independe
`Interim analysis of efficacy data: The s
`'
`after 50% and 75% of the expected numb
`'
`occurred. They plan t
`’
`difference in event rates
`analysis, and 3 standard deviations for the secon
`analysis 3 months later, the trial may be termma
`associated with these criteria is very sma
`primary analysis will be done at a 1 sided alpha=0.025.
`at the trial may also be stopped early ifthere is clear evidence of futility
`ating non—inferiority. At the time of the two formal interim analyses,
`The sponsor also notes th
`futility calculation of the conditional power to demonstrate non—
`with respect to demonstr
`e at the end of this trial. If in the
`there will be an interim
`lacebo on the primary outcom
`ower is unreasonably low (e.g. < 10%),
`they may
`inferiority of liraglutide versus p
`judgment of the DMC this conditional p
`recommend early stopping.
`
`terim analyses of efficacy data,
`cular outcome events have
`'
`boundary such that if the
`
`for the first interim
`
`alpha spending
`aluate the final
`
`'
`
`'
`
`'
`
`.
`
`Statistical review comments:
`
`the interim analyses of efiicacy will take place after 50%
`— Based on a total of 611 events,
`roximately 458) events have occurred.
`(approximately 306) and 75% (app
`maly and secondary cardiovascular,
`—
`”Eflicacy endpoints” in this study include the pri
`microvascular and all-cause mortality endpoints, and individual components of the4
`
`

`

`outcomes, as well as HbAlc, blood lipids, and other eflicacy endpoints. We would like the
`sponsor to clarify that which endpoints the Haybittle-Peto stopping boundary refers to.
`— The sponsor also plans to evaluate the eflicacy endpoints for futility after 50% and 75% of
`events have occurred. We would like the sponsor to clarifi} which endpoints will be
`evaluatedforfutility.
`— We also request clarification as to how this interim analysis will be conducted,
`maintain the appropriate study blind.
`
`in order to
`‘
`
`Statistical methods: The primary analysis will be a Cox regression including only treatment
`group as a covariate. Clinically relevant variables will be considered as covariates for further
`analyses that will be exploratory in nature. Major outcomes to be analyzed will be those that are
`confirmed by the Event Adjudication Committee (who will not have access to the treatment
`allocation at the time of adjudication) where applicable. All outcome analyses will be based on
`the time from randomization to the first occurrence of the outcome. Subjects who complete the
`study without having an outcome will be censored on the last day of their follow—up for the
`relevant analyses. After verifying the proportional hazards assumptions Visually, Cox models
`will be used to estimate the hazard ratios and 2—sided 95% confidence intervals and to calculate
`
`the P value. Non-inferiority 0f liraglutide vs. placebo will be assessed from the upper bound of
`the 2—sided 95% confidence interval, and by testing that the hazard ratio is significant less than
`1.3.
`If non—inferiority is established for the primary outcome, the data will then be tested for
`evidence of a significantly lower outcome hazard versus placebo. Additional exploratory
`analyses are also planned, as well as separate subgroup analyses based on gender, age group,
`body mass index, AlC, duration of diabetes, geographic region, and a history of a previous
`cardiovascular event.
`'
`
`Statistical review comment: The proposed statistical methods are acceptable.
`
`

`

`Application
`Type/Number
`
`Submission
`Type/Number
`
`Submitter Name
`
`Product Name
`
`NBA-22341
`
`ORlG-1
`
`NOVO NORDISK
`INC
`
`VICTOZA (LIRAGLUTIDE)
`
`This is a representation of an electronic record that was signed
`electronically and this page is the manifestation of the electronic
`signature.
`
`JANICE A DERR
`
`09/22/2009
`
`JON T SAHLROOT
`
`09/22/2009
`
`

`

`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 Number:
`
`022341/0
`
`Drug Name:
`
`VictozaTM (liraglutide) injection
`
`Indicati0n(s):
`
`Treatment of type 2 diabetes mellitus
`
`Applicant:
`
`Date(s):
`
`Novo Nordisk Inc
`
`Submission date: May 23, 2008
`
`PDUFA Goal Date (extended): May 23, 2009
`
`Advisory Committee Date: April 2, 2009
`
`Review Priority:
`
`Standard
`
`Biometrics Division:
`
`Division of Biometrics 2
`
`Statistical Reviewer:
`
`Janice Derr, PhD.
`
`Concurring Reviewers:
`
`J. Todd Sahlroot, Team Leader and Deputy Division Director
`
`Thomas J. Permutt, Director, Division of Biometrics 2
`
`Medical Division:
`
`Division of Metabolism and Endocrinology Products
`
`Clinical Team:
`
`Karen Mahoney, M.D., Medical Reviewer
`
`Lisa Yanoff, M.D., Medical Reviewer
`
`Hylton J offe, M.D., Medical Team Leader
`
`Mary H. Parks, M.D., Division Director
`
`Project Manager:
`
`John Bishai
`
`Keywords:
`
`clinical studies, NDA review
`
`

`

`Statistical review ofNDA 022341/0 Liraglutide for type 2 diabetes
`
`2/65
`
`Table of Contents
`
`1.
`
`EXECUTIVE SUMMARY ................................................................................................ 5
`
`Conclusions and Recommendations ............................................................................. 5
`1.1
`1.2 Brief Overview ofClinical Studies .............................................................................. 6
`
`1.3
`
`Statistical Issues and Findings ...................................................................................... 7
`
`2.
`
`INTRODUCTION .............................................................................................................. 8
`
`Overview ...................................................................................................................... 8
`2.1
`2.2 DataSources .......... 19
`
`3.
`
`STATISTICAL EVALUATION ...................................................................................... 20
`
`Evaluation of Efficacy ................................................................................................ 20
`3.1
`3.1 1. Subject disposition .....................................................................................................20
`3.1.2. Subject demographic and baseline characteristics ..................................................... 27
`3.1.3. Analysis populations ..................................-................................................................. 32
`3.1.4. Primary efficacy variable ............................................................................................ 32
`3.1.5. Statistical analysis methods for primary efficacy endpoint ........................................ 32
`3.1.6. Results ofthe statistical analysis ofefficacy.................................................. ........... 34
`3.1.7. Other Efficacy Endpoints: Body weight .................................................................... 40
`3.2
`Evaluation of Safety
`..............................................................48
`
`4.
`
`FINDINGS IN SPECIAL/SUBGROUP POPULATIONS .............................................48
`
`Gender, Race and Age ................................................................................................ 48
`4.1
`4.2 Other Special/Subgroup Populations ...............................-........................................... 53
`
`5.
`
`SUMMARY AND CONCLUSIONS ............................................................................... 56
`
`Statistical Issues and Collective Evidence .................................................................. 56
`5.1
`5.2 Conclusions ................................................................................................................ 56
`
`5.3
`
`Recommendations for Labeling .....
`
`................................. 56
`
`APPENDIX: PRIMARY EFFICACY ENDPOINT BY SUBGROUP CATEGORIES ..............58
`
`SIGNATURES/DISTRIBUTION LIST ........................................................................................ 65
`
`

`

`Statistical review of NDA 022341/0 Liraglutide for type 2 diabetes
`
`3/65
`
`TABLE]
`TABLEZ
`
`TABLE3
`TABLE4
`TABLES
`TABLE6
`
`TABLE7
`
`TABLE8
`TABLE9
`
`'TABLElO
`
`TABLEll
`
`TABLE12
`
`TABLE13
`
`TABLE14
`
`TABLE15
`
`TABLE16
`
`TABLE17
`
`TABLE18
`
`TABLE19
`
`LIST OF TABLES
`
`Overview of treatment regimens in the five therapeutic confirmatory trials ............. 10
`Number of randomized subjects and sites by country for each of the five
`Phase 3 studies ............................................................................................................ 1 1
`Data sources for studies .............................................................................................. 19
`
`Fasting Plasma Glucose criteria for the “ineffective therapy” classification .............20
`Subject disposition in each study23
`Subject demographic and baseline characteristics in the randomized subjects
`in each ofthe five key studies ....................................................................................28
`Baseline levels of HbAlc in randomized subjects in each of the five key
`studies (by arm) .............I............................................................................................. 30
`Analysis of HbAlc, change from baseline (LOCF, ITT analysis set) ....................... 37
`Comparison of active control comparator arm with placebo control arm,
`HbAlc primary endpoint (change from baseline), ANCOVA primary mode],
`LOCF/ICC primary analysis population .................................................................... 39
`Trial 1436; Mean changes from baseline in HbAlc (%) at endpoint: subgroup
`analysis of the ITT/LOCF population by age, gender and race ................................. 58
`Trial 1436; Mean changes from baseline in HbA]c (%) at endpoint: subgroup
`analysis of the ITT/LOCF population by baseline HbAlc category and
`baseline BMI category ................................................................................................59
`Trial 1572; Mean changes from baseline in HbA1c (%) at endpoint: subgroup
`analysis of the ITT/LOCF population by age, gender and race ................................. 60
`Trial 1572; Mean changes from baseline in HbA]c (%) at endpoint: subgroup
`analysis of the ITT/LOCF population by baseline HbAlc category and
`baseline BMI category ................................................................................................ 61
`Trial 1573; Mean changes from baseline in HbA1c (%) at endpoint: subgroup
`analysis of the ITT/LOCF population by age, gender, race and ethnicity .................62
`Trial 1573; Mean changes from baseline in HbA1c (%) at endpoint: subgroup
`analysis of the ITT/LOCF population by baseline HbAlc category and by
`baseline BMI category ................................................................................................ 62
`Trial 1574; Mean changes from baseline in HbA1c (%) at endpoint: subgroup
`analysis of the lTT/LOCF population by age, gender, race and ethnicity ................. 63
`Trial 1574; Mean changes from baseline in HbAIc (%) at endpoint: subgroup
`analysis of the ITT/LOCF population by baseline HbAlc category and by
`baseline BMI category ................................................................................................63
`Trial 1697; Mean changes from baseline in HbAlc (%) at endpoint: subgroup
`analysis of the ITT/LOCF population by age, gender and race ................................. 64
`Trial 1697; Mean changes from baseline in HbA}c (%) at endpoint: subgroup
`analysis of the ITT/LOCF population by baseline HbAlc category and by
`baseline BMI category ................................................................................................64
`
`

`

`Statistical review of NDA 022341/0 Liraglutide for type 2 diabetes
`
`4/65
`
`LIST OF FIGURES and EXHIBITS
`
`FIGURE 1
`FIGURE 2
`FIGURE 3
`FIGURE 4
`FIGURE 5
`FIGURE 6
`
`FIGURE 7
`FIGURE 8
`
`FIGURE 9
`FIGURE 10
`
`FIGURE 11
`FIGURE 12
`
`FIGURE 13
`FIGURE 14
`
`EXHIBIT 1
`EXHIBIT 2
`EXHIBIT 3
`EXHIBIT 4
`EXHIBIT 5
`
`Design ofTrial 1436 .................................................................................................. 13
`Design ofTrial 1572 .................................................................................................. 14
`Design ofTrial 1573 .................................................................................................. 15
`Design of Trial 1574 .................................................................................................. 16
`Design ofTrial 1697 .................................................................................................. 17
`Disposition by week on study; Kaplan—Meier plots (horizontal aXis shows the
`clinic visits where HbAlc was determined) ............................................................. 26
`Distribution of HbAlc at baseline ............................................................................. 31
`
`Forest plot of HbAlc, estimated mean difference i 95% Cl (LOCF, ITT
`analysis set) ............................................................................................................... 37
`Mean HbAlc over time in the five phase 3 studies ................................................... 38
`Forest plot of body weight (kg), estimated mean difference :1: 95% Cl
`(LOCF, ITT analysis set) .......................................................................................... 46
`Fasting Plasma Glucose over time in the five phase 3 studies ....................i.............47
`HbAlc primary efficacy endpoint in subgroups: Gender, age, race and
`ethnicity ..................................................................................................................... 50
`HbAlc primary efficacy endpoint, in Subgroups: Baseline HbAlc ......................... 54
`HbAlc primary efficacy endpoint, in subgroups: Baseline BMI ............................. 55
`
`Body weight at baseline and change from baseline in Trial 1436 ............................ 41
`Body weight at baseline and change from baseline in Trial 1572 ............................ 42
`Body weight at baseline and change from baseline in Trial 1573 ............................ 43
`Body weight at baseline and change from baseline in Trial 1574 ............................ 44
`Body weight at baseline and change from baseline in Trial 1697 ............................ 45
`
`

`

`Statistical review of NBA 022341/0 Liraglutide for type 2 diabetes
`
`5/65
`
`1.
`
`EXECUTIVE SUMMARY
`
`1.1
`
`Conclusions and Recommendations
`
`Efficacy Conclusions: Based on an evaluation of five key Phase 3 studies, I conclude that the
`efficacy of liraglutide 1.2 mg and 1.8 mg is supported by the comparisons to placebo and to
`active control comparators in a range of background antidiabetic therapies. The efficacy of
`liraglutide 0.6 mg is less well supported. The 0.6 mg dose is proposed for dose initiation, after
`which the dose levels of 1.2 mg or 1.8 mg may be selected, based on clinical response (see the
`draft patient insert under “Dosage and Administration.”
`
`Moao/flempy' Liraglutide monotherapy resulted in a net average reduction in HbAl c at week 52
`of 0.33 for the 1.2 mg dose and 0.62 for the 1.8 mg dose compared to the active comparator
`glimepiride 8 mg monotherapy. These comparisons were statistically significant in the direction
`of superiority of liraglutide monotherapy to the active control monotherapy.
`
`Add-012 Mempy- Liraglutide as an add-on therapy resulted in net average reductions in HbAlc at
`week 26 that ranged from 0.78 to 1.36 compared to placebo, with a range of background
`antidiabetic therapies, for the 1.2 mg dose and the 1.8 mg dose. These reductions were
`statistically significant
`in the direction of superiority to liraglutide add-on therapy.
`The
`background therapies were metformin 2 g, glimepiride 4 mg, metformin 2 g + rosiglitazone 8
`mg {4/7gfljflj, and glimepiride 4 mg + metformin 2 g.
`
`liraglutide compared to an active control resulted in
`With these same background therapies,
`either a non-inferior HbAlc response or superior HbAlc response, as summarized below:
`
`0 Liraglutide was non-inferior to glimepiride 4 mg for both the 1.2 mg dose and the 1.8 mg
`dose (metformin 2 g background therapy).
`
`0 Liraglutide was superior to rosiglitazone 4 mg, for both the 1.2 mg dose and the 1.8 mg
`dose (glimepiride 4 mg background therapy).
`Caveat:
`by trial design,
`the active
`comparator dose of rosiglitazone was one half the maximal FDA approved dose of 8 mg.
`
`0 Liraglutide 1.8 mg was superior to insulin glargine (glimepiride 4 mg + metformin 2 g
`background therapy); this statistical review does not address the adequacy of the glargine
`titration.
`
`The efficacy of liraglutide 0.6 mg is less well supported. Liraglutide 0.6 mg was non-inferior to
`rosiglitazone 0.4 mg (glimepiride background therapy). However, liraglutide 0.6 mg did not
`meet the criteria for non-inferiority to glimepiride 4 mg (metformin background therapy).
`
`

`

`Statistical review of NBA 022341/0 Liraglutide for type 2 diabetes
`
`6/65
`
`Although the studies were not powered for a comparison between liraglutide dose arms, and
`these comparisons were not included in the pre-specified sequential testing protocol, it can be
`noted that the 95% confidence intervals of the average HbAlc change from baseline for the 1.2
`mg and 1.8 mg dose arms overlapped to a great extent in three of the four studies in which both
`doses were evaluated.
`In the other study, the 95% confidence interval of the 1.8 mg dose arms
`overlapped less with the 95% C1 of the 1.2 mg dose,
`in the direction of a greater average
`reduction in l-lel c with the larger dose.
`
`Results for fasting plasma glucose supported the efficacy of liraglutide as monotherapy and as an
`add-on to background therapy with the other anti-diabetic drugs used in these studies.
`
`The average HbAlc response in the younger and older age groups (< 65 and Z 65 years) and in
`males and females were relatively similar. Most subjects were Caucasian in each of the five key
`studies.
`In the two studies with subjects from the US, the numbers of subjects in the other
`identified race categories were small and did not support an evaluation of potential race-related
`difference in HbAlc reduction.
`These two studies had reasonable representation in the
`Hispanic/Latino ethnicity subgroup, and the average HbAlc response was relatively similar in
`this subgroup compared to the non-Hispanic/Latino subgroup.
`
`The results from the phase 3 studies support the conclusion that liraglutide is associated with an
`average net loss in weight at 26 weeks and 52 weeks compared to several of the background
`diabetic therapies used in the studies. This may be a clinically relevant finding, considering that
`a range of 43% to 74% of subjects in the five phase 3 studies were classified as obese at baseline
`with a BM] 2 30 kg/mz. Approximately half of the subjects (ranging from 40% to 62%) in the
`liraglutide arms lost from 0% to 5% of their baseline body weight at the study endpoint.
`
`Conclusions regarding the safety of liraglutide are addressed in the
`Safety Conclusions:
`clinical
`review by Dr. Karen M. Mahoney and in the briefing document,
`“A Joint
`Clinical/Statistical Review of Cardiovascular Events and Thyroid Tumors,” by Dr. Mahoney and
`this reviewer, for the April 2, 2009, meeting of the Endocrinologic and Metabolic Drugs
`Advisory Committee.
`‘
`
`Recommendations: General recommendations for labeling are included in part 5.3 of this
`review.
`
`1.2
`
`Brief Overview of Clinical Studies
`
`The clinical development ofliraglutide included efficacy studies of liraglutide monotherapy and.
`add—on combination therapy with other common oral anti-diabetic drugs
`(glimepiride,
`metformin, rosiglitazone or insulin). Two general populations of subjects with type 2 diabetes
`mellitus were examined. A monotherapy phase 3 study was performed in subjects who had
`never
`received pharmacologic therapy or had received only minimal
`therapy.
`Add—on
`combination therapy studies were conducted in subjects who were inadequately controlled by
`
`

`

`Statistical review of NDA 022341/0 Liraglutide for type 2 diabetes
`
`7/65
`
`their existing therapy. The design of the five Phase 3 studies shared some common features and
`also had some differences. All studies were randomized, controlled and double-blind. The
`monotherapy study had an active control comparator arm, and the primary efficacy endpoint was
`evaluated after 52 weeks of treatment. The four add-on studies were evaluated after 26 weeks of
`
`Three of the add—on studies included both a placebo comparator arm and an active
`treatment.
`control comparator arm. One of the add-on studies had a placebo control arm but not an active
`control arm. Three dose levels ofliraglutide, 0.6 mg, 1.2 mg and 1.8 mg were evaluated in the
`Phase 3 program, but not all three arms were included in each study.
`
`The primary efficacy criterion in all major studies was the change from baseline to study
`endpoint (week 26 or 52) in glycated hemoglobin (HbAlc). Change in body weight was a key
`secondary efficacy endpoint. A total of 3992 subjects were randomized in five Phase 3 clinical
`studies. These five key studies are the focus of this statistical review.
`'
`
`1.3
`
`Statistical Issues and Findings
`
`Based on an evaluation of the five key Phase 3 studies, I conclude that the efficacy of liraglutide
`1.2 mg and 1.8 mg is supported by the comparisons to placebo and to active control comparators
`in a range of background antidiabetic therapies. The efficacy of liraglutide 0.6 mg is less well
`supported. The estimated effects of liraglutide on HbAlc change from baseline at week 26 and
`week 52 in the different
`target populations and background antidiabetic therapies are
`summarized in TABLE 8. The sponsor proposes to market all three doses; however, based on the
`“Indications and Use” section of the draft patient insert, the 0.6 mg dose level may be used for
`dose initiation, followed by an increase after at least one week at 0.6 mg to the 1.2 mg dose level.
`The 1.8 mg dose may be used after at least one week on the 1.2 mg dose, “.———’——+
`4—..-
`” (draft patient insert).
`'
`
`[1(4)
`
`Support for the efficacy of liraglutide compared to a placebo control and compared to an active
`control also comes from a consistent pattern of early withdrawals due to ineffective therapy,
`when observed across the five studies.
`In the four studies that had a placebo add-on arm,
`subjects in this arm were more likely to withdraw early due to ineffective therapy than subjects
`in the liraglutide arms. In the four studies that had an active comparator arm, subjects in this arm
`were about equally likely to withdraw early due to ineffective therapy as subjects in the
`liraglutide arms.
`
`A potential concern for the statistical analysis of the primary endpoint arose for the monotherapy
`study, because of the occurrence of a substantial percentage of subjects with HbAlc S 7.0 at
`baseline.
`In the monotherapy study, conducted with an active control comparator, 11.7% of
`subjects had baseline HbAlc levels 5 7.0, and another 18.1% of subjects had baseline HbAlc
`between 7.0 and 7.5. This relatively high proportion of subjects who were in reasonable diabetic
`control at baseline raised the concern that both the active control comparator and the liraglutide _
`arms would tend to have a'small average change from baseline HbAlc at the study endpoint.
`This assumption comes from a general finding across clinical studies of anti-diabetic drugs that
`
`

`

`Statistical review of NDA 022341/0 Liraglutide for type 2 diabetes
`
`8/65
`
`subjects with lower levels of HbAIc at baseline tend to experience smaller decreases in HbAlc
`at the study endpoint compared to subjects with higher levels at baseline.
`In this situation, the
`assay sensitivity of the comparison may not have supported a non-inferiority margin of 0.4.
`However,
`the two liraglutide arms were superior to the active control arm for the primary
`endpoint, with statistically significant differences for both comparisons. For this reason, the
`proportion of subjects in reasonable diabetic control at baseline was not a review issue.
`However, this topic is an important consideration for future active—controlled studies.
`
`The inclusion of both an active control arm and a placebo control arm in three of the studies
`presented an opportunity to estimate the placebo—adjusted effect of the active control comparator
`within the study.
`In all three studies, the placebo-adjusted effect was statistically significantly
`different from 0. The net effect of glimepiride was similar to the results from the three historica