`Amneal Pharmaceuticals LLC v. Alkermes Pharma Ireland Limited
`IPR2018-00943
`
`Page 1 of 9
`
`
`
`alcohol dependence”as part of an appropriate plan of managementfor addictions.
`Naltrexone has not been widely used forthis indication dueto the general beliefthatits
`efficacyis limited, and that poor complianceis one of the more significant factors
`contributing to this limited efficacy. The sponsor has proposed that an extended-release
`depot preparation may improve complianceand, therefore, effectiveness. They have also
`proposedthat the absence ofa first-pass effect in the liver may decrease the hepatic
`toxicity noted in the original naltrexone application resulting in the inclusion of a boxed
`warningin the packageinsert.
`.
`
`Review of the CMCportion of this application was completed by Jila H. Boal, Ph.D.
`Review of the pharmacology and toxicology data presented in this application was
`completed by Mamata De, Ph.D. A supervisory review was provided by Daniel Mellon,
`Ph.D., Supervisory Pharmacologist in this division. Review ofthe clinical pharmacology
`and biopharmaceutics data in the application was completed by Srikanth C. Nallani,
`Ph.D. A clinical review of the safety and efficacy data submitted was completed by
`Mwango Kashoki, M.D., M.P.H.Astatistical review and evaluation was completed by
`Dionne Price, Ph.D. Celia Winchell, M.D. provided a supervisory review ofthe
`application. Consultation on this application was also obtained from the Division of
`Pulmonary and Allergy Products (DPAP), the Division of Drug Marketing, Advertising
`and Communications (DDMAC), and the Office of Drug Safety (ODS).
`
`Asthe clinical and statistical reviews have thoroughly detailed and analyzed the data
`submitted in this application, I will only briefly summarizetheir findings in this memo.
`
`Efficacy:
`
`A single adequate and well-controlled study was submitted in support ofefficacy. Study
`21-003 (003) was a multicenter, randomized, placebo-controlled, double-blind, parallel-
`group study comparing Vivitrol™ (190 mg or 380 mg) and placebofor six months.
`Adults meeting the DSM IV diagnostic criteria for alcohol dependence, and who hadat
`least two episodes of heavy drinking (4 drinks per day for womenand5 drinks per day for
`men) per week were admitted to the study. Complete abstinenceat baseline was not
`required. Subjects received monthly intramuscularinjection of drug or placeboin the
`gluteal muscle.
`
`Alcohol consumption wascollected using the Time Line Follow-back Method and the
`quantity then converted into a numberofstandard drinks using a protocol-specified
`definition/formula. Psychosocial treatment was provided using the BRENDA
`(Biopsychosocial, Report, Empathy, Needs, Direct advice and Assessment of
`responsiveness) model. The protocol-specified primary outcome analysis was a
`comparison of the event rate of heavy drinking with heavy drinking definedas at least
`four drinks per day for womenandfive drinks a day for men.
`
`NDA 21-897 Division Director’s Approvable Memo
`Vivitrol™
`
`2
`
`December23, 2005
`
`Page 2 of 9
`
`Page 2 of 9
`
`
`
`Recent analyses conducted by the NIAAA documented an apparent link between various
`patterns of drinking andthe likelihood of drinking-related psychosocial consequences.
`Theresults of these analyses suggest that the strongest predictor of avoiding significant
`consequencesis the absence of any heavy drinking days (employing observation periods
`of 3 to 12 months), with heavy drinking days defined as more than four drinks for males
`and morethan three drinks for females. Therefore, at the request of the Division, a
`responder analysis was performedto add perspective onthe clinical relevance of the
`results of the primary analysis. The agreed upon respondercategories included:
`
`no heavy drinking days per month
`0 and < 1 heavy drinking day per month
`1 and <2 heavydrinking days per month
`2 and <3 heavy drinking days per month
`3 and < 4 heavydrinking days per month
`4 heavy drinking days per month
`
`The results of the primary outcome analysis demonstrateda statistically significant
`treatment effect for the 380-mg dose only. Dr. Price’s Table 5 summarizingthis data is
`reproduced below:
`
`Comparison of Median Eventrate of Heavy Drinking: Non-Parametric Analyses
`
`-Any missing data day is defined as a heavy drinking day
`p-value’
`Wilcoxontest
`unstratified
`
`Treatment Group
`
`N
`
`204
`Placebo
`206
`190mg
`201
`380 m
`p-value compared to placebo
`
`-
`
`Median Event Rate of
`Heavy Drinking
`0.35
`0.30
`0.20
`
`Percent
`Difference
`
`13%
`41%
`
`0.69
`0.05
`
`The sponsor also analyzed the data based on abstinence at baseline (defined as abstinent
`for 7 days prior to treatment) and based on subjects’ treatment goalat baseline (total
`abstinence or several other options). While the subjects’ treatment goal did not appear to
`influence the outcome, whether or not a subject was abstinent at baseline had a profound
`effect on the subject’s response to treatment. The data supporting this conclusionis
`summarized in Dr. Winchell’s table from page 12 of her review, reproduced below:
`
`HAZARD RATIO(P-VALUE)
`|__| NUMBEROF SUBJECTS
`
`PLACEBO 190 MG 380MG|190 MG VS. 380 MG VS.
`FACTOR
`PLACEBO
`PLACEBO
`188
`193
`190
`Yes
`Lead-in
`0.925 (0.4803)
`0.790 (0.0532)
`17
`17
`19
`No
`Drinking
`0.049 (<0.0001)
`0.202 (0.0053)
`
`Treatment Goal|Yes 90 90 90 0.879 (0.4994) 0.718 (0.1119)
`
`
`
`
`
`
`of Abstinence|No 119 120 115 0.912 (0.4841) 0.785 (0.0991)
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`NDA21-897 Division Director’s Approvable Memo
`Vivitrol™
`
`3
`
`December23, 2005
`
`Page 3 of 9
`
`Page 3 of 9
`
`
`
`Theresults of the responder analysis showed a smalleffect of treatment and only at
`greater than 1 heavy drinking day per month. However, whenthe effect of abstinence at
`baseline was includedin the analysis, a much largereffect was seen for all strata,
`including 0 heavy drinking days per month. The data supporting these conclusions are
`summarized in Dr. Winchell’s tables from pages 13 and 14 of her review, reproduced
`below:
`
`Responderanalysis using 5/4 definition of responders
`and 2-month grace period.
`HDDper
`Placebo
`190 mg
`380 mg
`
`month
`(n=204)
`(n=206)
`(n=201)
`0
`22 (11%)
`25 (12%)
`26 (13%)
`0-1
`36 (18%)
`37 (18%)
`39 (19%)
`0-2
`47 (23%)
`51 (25%)
`61 (30%)
`0-3
`52 (26%)
`59 (29%)
`70 (35%)
`
`0-4 79 (39%) 56 (28%) 65 (32%)
`
`
`
`Responderanalysis using 5/4 definition of responders and 2-month grace period.
`Placebo-
`190 mg
`380 mg
`
`HDDper
`Non-
`Abstinent
`Non-
`Abstinent
`Non-
`Abstinent
`month
`abstinent
`abstinent
`abstinent
`
`(n = 186)
`(n=18)
`(n= 189)
`(n=17)
`(n = 184)
`(n=17)
`20 (11%)
`2 (11%)
`15 (8%)
`10 (59%)
`19 (10%)
`7 (41%)
`0
`31 (17%)
`5 (28%)
`27 (14%)
`10 (59%)
`30 (16%)
`9 (53%)
`0-1
`40 (22%)
`7 (39%)
`41 (22%)
`10 (59%)
`49 (27%)
`12 (71%)
`0-2
`44 (24%)
`8 (44%)
`49 (26%)
`10 (59%)
`58 (32%)
`12 (71%)
`0-3
`
` 0-4 48 (26%) 8 (44%) 55 (29%) 10 (59%) 65 (35%) 14 (82%)
`
`
`
`
`
`
`Clinical Safety:
`
`Exposure
`
`Overone thousand subjects were exposed to Vivitrol™. Dr. Winchell’s summary table
`of exposure by numberofinjections (page 16 of her review) is reproduced below:
`
`
`
`
`
`541
`
`394
`127
`
`Atleast 1 injection
`Atleast 3 injections
`
`<380 mg
`349
`
`217
`177
`98
`
`
`
`
`
`
`
`
`
`22
`
`
`NDA21-897 Division Director’s Approvable Memo
`Vivitrol™
`
`4
`
`December23, 2005
`
`Page 4 of 9
`
`Page 4 of 9
`
`
`
`Deaths
`
`Five deaths occurred in the Vivitrol™ database. Based on Drs. Kashoki and Winchell’s
`reviews, only two of those deaths were possibly related to study drug exposure. These
`two deaths were both suicides in subjects treated with study drug for extendedperiods of
`time. Oneoccurred after five months of treatment, but not until two months after the last
`dose. The other occurred after the subject had received 33 doses.
`
`Discontinuations Due to Adverse Events (AEs)
`
`-
`
`There wasa slightly higher rate of dropout due to adverse events for the study drug-
`treated subjects compared to the placebo-treated subjects. However, there was no clear
`dose effect. The most commonreasonsfor discontinuation were:
`injection site reactions,
`alcoholism (i.e., lack of efficacy), nausea, pregnancy, abnormal LFTs,and suicide-related
`AEs. There wasa slightly higher incidence of dropout dueto suicidal behavior for the
`drug-treated vs. the placebo-treated subjects, 0.9% vs. 0%, respectively). There was also
`a slightly higher incidence of dropout for depression, 0.3% vs. 0% for the drug vs.
`placebo-treated subjects, respectively. Neither of these events appearedto be dose-
`related, and the percentage of subjects dropping out for depression washighest in subjects
`treated with oral naltrexone.
`
`Serious Adverse Events
`
`Suicide-related serious AEs were reported more frequently in the drug-treated subjects
`compared to the placebo-treated subjects (1.4% vs. 0%, respectively). One subjectin the
`380-mg treatment group developed a severe injection site reaction described as necrosis
`requiring fairly extensive tissue excision. Histopathological evaluation of the excised
`tissue documented a “hypersensitivity reaction.” One subject treated with 380-mg
`Vivitrol™ developed apparent eosinophilic pneumonianot responsiveto antibiotics, but
`responsive to steroid treatment.
`
`Common Adverse Events
`
`The following gastrointestinal adverse events occurred more frequently in the Vivitrol™-
`treated subjects: nausea, vomiting, diarrhea, abdominal pain, dry mouth,
`flatulence/bloating, decreased appetite and decreased weight. Additional adverse events
`that occurred with greater frequency in Vivitrol™-treated subjects were: asthenia,
`injection site reactions, headache, dizziness, somnolence/sedation, muscle cramps,
`arthralgia, back pain, rash, angioedema/urticaria, anxiety, and depression and/or suicidal
`ideation.
`
`While abnormal LFTs occurred with slightly greater frequency in the drug-treated
`subjects, the rates were comparablefor the Vivitrol™-treated subjects and theoral
`naltrexone-treated subjects. Injection site reactions in the placebo-treated subjects were
`
`NDA 21-897 Division Director’s Approvable Memo
`Vivitrol™
`
`December 23, 2005
`
`5
`
`Page 5 of 9
`
`Page 5 of 9
`
`
`
`generally innocuous tenderness, while induration andpruritis were seen commonlyin the
`Vivitrol™-treated subjects. Injection site pain was seen most often in the higher dose
`group, suggesting that naltrexoneitself is serving as an irritant. Depression also appeared
`to occur abouttwiceas frequently in the drug-treated subjects comparedto the placebo-
`treated subjects.
`
`Ofnote, elevated eosinophil counts occurred with greater frequency in Vivitrol™-treated
`subjects and with the extent of elevation occurring in an apparently dose-related pattern.
`Additionally, all twelve cases ofurticaria and angioedemaoccurred in Vivitrol™-treated
`subjects.
`
`Medication Errors:
`
`The Division of Medication Errors and Technical Support (DMETS)in the Office of
`Drug Safety has recommendedthat the Vivitrol™ Kit not contain the proposed three
`syringe needles(i.e., two 20-gauge 1 % inch and one 20-gauge % inch), as “This may
`cause confusion anderror as healthcare practitioners may inadvertently use the 1 % inch
`needle for reconstitution and then switchto the shorter % inch needlefor the
`intramuscular (IM) injection. Additionally, some practitioners may not switch the
`needlesprior to administration.”
`
`I do not agree with this speculative scenario. Physicians, nurses and other health-care
`practitioners are quite familiar with the need to use a longer needle for a gluteal IM
`injection. The longer needle would also maketransfer ofthe diluent moredifficult.
`
`Nonclinical Safety:
`
`Dr. De has recommended thatthis application should not be approvedatthis time due to
`the absence of adequate evidence that the exposure (toxicokinetic data) in the referenced
`naltrexonepreclinical studies provides support for the higher exposures foundin the
`clinical pharmacokinetic studies for Vivitrol™ compared to the oral formulation, and the
`consequent need for the sponsor to perform SegmentsI, II and III reproductive toxicity
`studies and carcinogenicity studies in two species. However, Dr. Mellon has
`recommendedthat the application is approvable. While he concurs with Dr. Dethat there
`is currently inadequate preclinical support for the naltrexone exposure levels found with
`Vivitrol™, he has concluded that the sponsor maybeable toperform a bridging study
`that will allow interpretation of the relative exposure to naltrexone betweenthe existing
`animal and humanstudies, thereby obviating the need for additional toxicology studies.
`If the sponsoris unable to document adequatepreclinical support for the higher exposure
`levels based on this bridging study, he recommendsthat the reproductive toxicology
`studies and the carcinogenicity studies would then be required.
`
`NDA21-897 Division Director’s Approvable Memo
`Vivitro]™
`‘December 23, 2005
`
`6
`
`Page 6 of 9
`
`Page 6 of 9
`
`
`
`Dr. Mellon has also determined that the references to products other than Reviacited in
`this application are not necessary for a determinationofthe preclinical safety of
`Vivitrol™and, therefore, the absenceofpatent certification and relative bioavailability
`studies for these references is moot.
`
`Biopharmaceutics:
`
`Dr. Nallani has concluded that the application is approvable if the sponsor agrees to
`provide revisionsto the drug release specifications to include the addition of appropriate
`Day 14 and Day 28 drug-release information.
`In addition, he recommendsthat the
`sponsorshould be required to agree to the following Phase 4 commitments:
`
`¢
`
`*
`
`conduct in vitro CYP inhibition studies using conventional substrates, as the data
`submitted in the application were drawn from studies employing fluorescent
`substrates whichtend to introduce non-specificity in detection, and;
`
`in vitro studies in human hepatocytes to evaluate the potential of
`conduct
`naltrexone to induce CYP3A4 and CYP1A2.
`
`is important to note that Dr. Nallani has also determined that Vivitrol™ has, on
`It
`average, a 4-fold greater AUC thanthe oral formulation of naltrexone.
`
`Chemistry, Manufacturing and Controls:
`
`Dr. Boal has concludedthat the application is approvable based on the CMC data
`submitted, but that the sponsor should agree to the following Phase 4 commitment:
`
`¢ Assessthe in vitro drug release data and percent crystallinity for the first five
`commercial batchesin order to tighten the ranges for the in vitro drug release
`specifications and —_
`the percent crystallinity of
`naltrexonein the microspheres.
`
`Discussion:
`
`The sponsorhas provided evidence that Vivitrol™ is effective for the treatment of
`alcohol dependence,butonly in patients who are abstinent for seven daysat theinitiation
`of treatment. While there was a numericaltrend supportive of non-abstinent subjects
`being responsiveto treatment with Vivitrol™, the overwhelmingsourceof the
`statistically significant treatment effect found in the primary outcome analysis came from
`the abstinent-at-baseline subjects. This finding was also supported by the responder
`analyses.
`
`NDA 21-897 Division Director’s Approvable Memo
`Vivitrol™
`December23, 2005
`
`.
`
`7
`
`Page 7 of 9
`
`Page 7 of 9
`
`
`
`Vivitrol™appears to havea significantly more concerning adverse eventprofile
`comparedto the approved ora] formulation of naltrexone. The most concerning set of
`adverse events that appear to be unique to this formulation are those related to the
`immunesystem: a notably high frequency of peripheral eosinophilia and frequent skin
`reactions (one quite serious, requiring extensive tissue excision) in Vivitrol™-treated
`subjects; twelve cases of urticaria and angioedemaoccurring onlyin Vivitrol™-treated
`subjects; and one case of apparent eosinophilic pneumonia in a subject treated with
`Vivitrol™,
`I concur with our own expert consultants from DPAPthat these findings do
`not represent clear evidence of a specific immunologic effect. However, the presence of
`all of these abnormalities, especially the presence of the notable case of probable
`eosinophilic pneumonia, an extremely rare and life-threatening disorder when nottreated
`quickly and appropriately, has raised a high level of concern regardingthe safety ofthis
`product.
`
`In light of these safety concerns, we must considerthe risk:benefit ratio for the to-be-
`treated patient population. As noted above,the likelihood of achieving effective
`treatment with Vivitrol™appears to be differentially related to drinking status at the
`initiation of treatment. Abstinent patients have a relatively high degree of success and,
`thus, the benefits associated with treatment would likely outweigh the risks associated
`with untreated alcoholism. However,it is unclear whether Vivitrol™is truly effective in
`
`non-abstinentpatients
`
`fof fry
`
`In additionto the clinical safety concerns noted above, the sponsorhas not provided
`adequate preclinical support for the naltrexone exposure levels found with Vivitrol™.
`Dueto the fact that Vivitrol™results in a 4-fold higher exposure to naltrexone compared
`to the approved oral formulation, it will be necessary for the sponsor to provide data from
`a bridging toxicokinetic study that will allow interpretation of the relative exposure to
`naltrexone based on the currently existing animal and humanstudies, thereby obviating
`the need for additional toxicology studies. If, however, the sponsor is unable to document
`adequatepreclinical support for the higher exposurelevels based on this bridging study,
`reproductive toxicology studies and carcinogenicity studies would then be required.
`
`Action:
`
`Approvable
`
`Bob A. Rappaport, M.D.
`Director
`Division of Anesthesia, Analgesia and Rheumatology Products
`- Office of Drug Evaluation II, CDER, FDA
`
`NDA 21-897 Division Director’s Approvable Memo
`:
`Vivitrol™
`December23, 2005
`
`8
`
`Page 8 of 9
`
`Page 8 of 9
`
`
`
`This is a representation of an electronic record that was signed electronically and
`this page is the manifestation of the electronic signature.
`
`Bob Rappaport
`12/23/2005 04:06:31 PM
`MEDICAL OFFICER
`
`Page 9 of 9
`
`Page 9 of 9
`
`