ALKERMES EXHIBIT 2013
`Amneal Pharmaceuticals LLC v. Alkermes Pharma Ireland Limited
`IPR2018-00943
`
`Page 1 of 16
`
`

`

`Docket No. FDA-2007—D—03 69
`
`June 2, 2014
`Page 2
`
`from alcohol in an outpatient setting prior to initiation of treatment, and for the prevention of
`relapse to opioid dependence following opioid detoxification. Vivitrol should be administered as
`
`part of a comprehensive management program that includes psychosocial support. The product
`is designed to deliver therapeutic levels of naltrexone both quickly and on a sustained basis, to
`help patients avoid relapse during the early stages of their therapy and then to provide
`maintenance levels of naltrexone over a full, 30—day dosing cycle.
`
`To achieve the release profile on which FDA’s approval is based, Vivitrol relies on a
`precise formulation of polylactide-co-glycolide (PLG) microspheres, as well as a delivery system
`that deposits the microspheres locally at the IM gluteal injection site. From the local depot site,
`the release of the drug is characterized by an initial peak approximately two hours after injection,
`followed by a second peak approximately two to three days after injection. Plasma concentration
`levels then begin a slow decline at or about day 14. The release of the drug is dependent on,
`among other factors, the specific composition and formulation of the microspheres and the
`accompanying diluent, manufacturing and product quality, and the dynamic interaction between
`the microspheres and the conditions at the IM injection and depot site.
`
`Alkermes appreciates this opportunity to comment on the 2014 Draft Guidance. The
`Company recognizes that in this newly revised version of the guidance, the Agency has made the
`significant addition of recommending that generic sponsors conduct an in vitro drug release
`study. Alkermes supports the addition of this new study. However, further amendment of the
`
`2014 Draft Guidance is needed to assure that any proposed generic products match Vivitrol’s
`
`proven safety and efficacy profile.
`
`EXECUTIVE SUMMARY
`
`Vivitrol is a long-acting, modified release inj ectable drug product that delivers its active
`ingredient in three clinically significant phases:
`
`0 An “initial phase,” during which naltrexone is immediately absorbed from the
`
`surface of the formulation’s microspheres into the systemic circulation;
`
`o A “hydration phase,” during which physical erosion of the microspheres begins,
`triggering the release of subsurface naltrexone; and
`
`o A “sustained release phase,” during which the microspheres steadily erode and
`
`allow a constant amount of drug to be absorbed.
`
`The multiphasic release of naltrexone from the Vivitrol microspheres ensures that
`circulating naltrexone levels are quickly raised to a therapeutic level and then sustained over the
`course of the month-long dosing interval. Vivitrol therefore strikes a careful balance between
`
`Page 2 of 16
`
`

`

`Docket No. FDA-2007—D—0369
`
`June 2, 2014
`Page 3
`
`to ensure therapeutic maintenance over the full, 30—day dosing cycle. Vivitrol is indicated for
`treatment of alcohol dependence in patients who are able to abstain from alcohol in an outpatient
`setting prior to initiation of treatment. Vivitrol is also indicated for the prevention of relapse to
`opioid dependence, following opioid detoxification.
`
`To help ensure that generic versions of Vivitrol are able to match the multiphasic release
`profile of Vivitrol, Alkermes believes that the 2014 Draft Guidance should be further revised in
`
`two key respects.
`
`First, consistent with Alkermes’ 2011 comments, the in vivo study included in the 2014
`Draft Guidance should be amended to include statistical analyses of additional pharmacokinetic
`(PK) parameters that are necessary to ensure equivalence to a multiphasic, modified release,
`depot product such as Vivitrol. In support of this request, Alkermes has completed a series of in
`silico (computer-based) simulations comparing Vivitrol against potential comparator
`formulations. These simulations found that the traditional in viva bioequivalence metrics (i. e.,
`Cmax, AUCo.t, and AUCOM) are inadequate and could lead to non-equivalent generic versions of
`Vivitrol passing the Agency’s proposed BE standards.
`
`Thus, to the extent the 2014 Draft Guidance relies solely on conventional BE measures,
`the Guidance is insufficient to ensure equivalence. Because matching Vivitrol’s in vivo release
`profile is critical for generic products, at least one additional PK parameter is needed to ensure
`true bioequivalence. As proposed herein, the 2014 Draft Guidance should be revised to include
`the partial AUC metric AUC10_23. This additional metric is able to detect significant differences
`in the segment of the concentration-time profile that is most susceptible to potential therapeutic
`equivalence deficiencies.
`
`Second, also consistent with Alkermes’ 2011 comments, FDA should revise the 2014
`
`Draft Guidance to ensure that no significant difference in injection site reactions would exist
`between Vivitrol and generic products. In particular, the Company asks the Agency to include
`an in vivo comparative study designed to assess local injection site safety for any proposed
`generic. This injection site study could be conducted as its own stand-alone study, or
`
`incorporated as part of the required in vivo PK study.
`
`We appreciate FDA’s continued efforts to develop guidance and to request public
`comment on recommended studies for generic drugs. In light of Alkermes’ experience with
`Vivitrol and with sustained release drug formulations, we submit the following comments on the
`2014 Guidance for the Agency’s consideration.
`
`Page 3 of 16
`
`

`

`Docket No. FDA-2007-D—0369
`
`June 2, 2014
`Page 4
`
`BACKGROUND
`
`A.
`
`
`Vivitrol
`
`As explained in Alkermes’ 2011 comment document, the Agency first approved Vivitrol
`on April 13, 2006, under new drug application (NDA) 21-897. Vivitrol is currently approved for
`(a) the treatment of alcohol dependence in patients who are able to abstain from alcohol in an
`outpatient setting prior to initiation of treatment, and (b) the prevention of relapse to opioid
`dependence, following opioid detoxification. In both cases, treatment with Vivitrol should be
`
`part of a comprehensive management program that includes psychosocial support.
`
`The active ingredient in Vivitrol, naltrexone, is microencapsulated in a high molecular
`weight 75/25 PLG polymer at a concentration of 337 mg of naltrexone per gram of microspheres.
`Physicians are instructed to administer the product as an IM gluteal injection, using only the
`needles provided, alternating sides for subsequent injections.4 The recommended dose is one
`injection every four weeks or once a month.
`
`Vivitrol is a long-acting, modified release inj ectable drug product that delivers its active
`ingredient in the following three clinically significant phases: an initial phase, during which
`naltrexone is immediately absorbed from the surface of the formulation’s microspheres into the
`systemic circulation; a hydration phase, during which physical erosion of the microspheres
`begins, triggering the release of subsurface naltrexone; and a sustained release phase, during
`which the microspheres steadily erode and allow a constant amount of drug to be absorbed. The
`
`extent of naltrexone released at each of these three clinically important phases is a carefully
`
`controlled manufacturing and formulation-dependent variable.
`
`B.
`
`In Vivo Bioeguivalence For IM Depot Products
`
`With regard to the in vivo bioequivalence of systemically absorbed drugs, FDA generally
`
`recommends that applicants conduct single dose crossover or parallel group PK studies,
`measuring the concentrations of the generic and reference drugs in the blood or plasma.
`Applicants then calculate a variety of measurements of the resulting drug concentration versus
`
`time curve, including (1) the maximum concentration, Cmax; (2) the time to maximum
`
`concentration, TmaX; and (3) the area under the curve (AUC), both from dosing until the last
`
`measured time point, AUCo_t, and extrapolated to infinity, AUC0_00. Cmax and Tmax are believed to
`
`reflect the rate of the drug’s absorption into the systemic circulation, and the AUC measures are
`
`believed to reflect the extent of the drug’s absorption.
`
`4 Different length administration needles are provided to accommodate varying body physiques, given the
`
`Page 4 of 16
`
`

`

`Docket No. FDA-2007-D-O369
`
`June 2, 2014
`Page 5
`
`FDA has acknowledged that there are certain drugs for which the statistical comparison
`of Cmax and AUC as described above are insufficient to ensure that generic products will be as
`safe and effective as the RLD.5 This is particularly so for products that are approved and labeled
`for extended-release and whose release properties depend on complex formulations designed to
`release drug over multiple phases.6 For example, FDA has recommended partial AUC analysis
`for products such as Metadate CD, Concerta, Ritalin LA, Adderall XR, Focalin XR, and Ambien
`CR.7 In citizen petition responses regarding these products, FDA has stated it will require as a
`condition of ANDA approval that the generic product be BE to the RLD at the 90% confidence
`
`interval using appropriately tailored partial AUC metrics in addition to the conventional metrics
`Cm... AUCO. and AUco-mf.8
`
`Products such as Vivitrol — long acting depot formulations that are injected, and that
`reside subcutaneously over the course of the dosing interval, releasing drug in a multiphasic and
`sustained fashion — are a particularly complex extended—release dosage form, and present a new
`and challenging BE problem. In the past 12 months, FDA has released revised BE guidance
`documents for several of these products. These revised guidances reflect FDA’s ongoing process
`of addressing the BE challenges such products pose. Collectively, these BE guidances
`underscore the general principle that additional measures should be required for such products,
`but that there is no “one—size—fits—all” approach.
`
`For example, in February 2014, in addition to the revised guidance for Vivitrol, FDA
`issued revised draft BE guidances for Trelstar (triptorelin pamoate for inj ectable suspension),
`Lupron Depot (leuprolide acetate for depot suspension), and Sandostatin LAR Depot (octreotide
`
`5 See, e. g., FDA Citizen Petition Response, Daytrana, Docket No. FDA-2012—P-0932 (Jan. 23, 2013) at 3 (“FDA
`recognizes that, under certain circumstances, it may be appropriate to use a partial AUC parameter to ensure
`comparable therapeutic effects”).
`
`6 See generally Briefing Package for the April 13, 2010 Meeting of the Advisory Committee for Pharmaceutical
`Science and Clinical Pharmacology, available at:
`hm;://www.fda.gov/AdvisogCommittees/CommitteesMeetingMaterials/Drugs/AdvisoryCommitteeforPharmaceutic
`alScienceandClinicalPharmacology/ucmZO1700.htm.
`
`7 See Draft Guidance on Methylphenidate Hydrochloride (Metadate CD) (Sept. 2012); Draft Guidance on
`Methylphenidate Hydrochloride (Concerta) (Sept. 2012); Draft Guidance on Methylphenidate Hydrochloride
`(Ritalin LA) (Nov. 2011); Draft Guidance on Amphetamine Sulfate; Dextroamphetamine Saccharate;
`Dextroamphetamine Sulfate (Adderall) (Sept. 2012); Draft Guidance on Dexmethylphenidate Hydrochloride
`(Focalin XR) (Mar. 2012); Guidance on Zolpidem (Ambien CR) (Final) (Oct. 2011). Available at
`hm;://www.fda.gov/Drugs/GuidanceComplianceRegulatogInformation/Guidances/ucm075207.htrn.
`
`8 See FDA Citizen Petition Response, Metadate CD and Concerta, Docket Nos. FDA-2004-P-0151 and FDA-2004-
`P-0290 (July 19, 2012); FDA Citizen Petition Response, Adderall XR, Docket No. FDA-2005-P-0120 (June 22,
`2012); FDA Citizen Petition Response, Ambien CR, Docket No. FDA-2007~P-0182 (Oct. 13, 2010). FDA has also
`required the use of partial AUC metrics in cases where the RLD displays time-dependent localization of absorption
`within the gastrointestinal tract. See FDA Citizen Petition Response, Asacol, Asacol HD and Pentasa, Docket Nos.
`
`Page 5 of 16
`
`

`

`Docket No. FDA-2007-D-0369
`
`June 2, 2014
`Page 6
`
`acetate for inj ectable suspension).9 In each of these cases, FDA has recognized that conventional
`analysis of in vivo PK data is insufficient to assure BE.
`
`In particular, for Trelstar, Lupron Depot, and Sandostatin LAR Depot, FDA has
`specifically recognized the need for partial AUC analysis. The dosing and release features of
`these depot products, along with the key aspects of FDA’s revised BE recommendations, are as
`follows:
`
`0 Trelstar is available in three strengths: 3.75 mg, 11.25 mg, and 22.5 mg. The
`product is administered through a single intramuscular injection, with a dosing
`interval of 4 weeks, 12 weeks, and 24 weeks respectively. In the revised
`guidance, FDA recommended a single-dose, in vivo PK study at each strength,
`
`with BE based on AUC“, in addition to conventional analysis of AUCo_t, AUC0_w,
`
`and Cmax, where AUC“ is the area under the PK curve from day 7 to the last
`
`sampling time point of the dosing interval.
`
`0 Lupron Depot is available in strengths of 3.75 mg, 7.5 mg, 11.75 mg, 15 mg, 22.5
`mg, 30 mg, and 45 mg, administered through intramuscular injection, with dosing
`intervals ranging from 1 month to 6 months. Again, FDA recommended single—
`dose, in vivo PK studies, with BB based on AUC“, in addition to the conventional
`metrics.
`
`o
`
`Sandostatin LAR Depot is available in strengths of 10 mg, 20 mg, and 30 mg, and
`is administered intragluteally every 4 weeks (10 mg and 20 mg) or every 8 weeks
`(30 mg). Here, FDA has recommended a single-dose, in vivo PK study at 30 mg
`with BE based on AUC0.23 and AUC28_56, in addition to the conventional
`measurements.
`
`In each of these cases, FDA has recognized that the features of these depot products
`require additional analysis to ensure therapeutic equivalency. Specifically, partial AUC tailored
`to the PK of the product is an integral part of demonstrating BE across the length of the dosing
`interval.
`
`Similarly, FDA issued revised BE recommendations for Invega Sustenna (paliperidone
`palmitate extended—release inj ectable suspension) in December 2013, and Risperdal Consta
`
`9 Draft Guidance on Triptorelin Pamoate (Trelstar) (Feb. 2014); Draft Guidance on Leuprolide Acetate (Lupron
`Depot) (3.75 mg, 7.5 mg, 11.25 mg, 15 mg, 22.5 mg) (Feb. 2014); Draft Guidance on Leuprolide Acetate (Lupron
`Depot) (30 mg, 45 mg) (Feb. 2014); Draft Guidance on Octreotide Acetate (Sandostatin LAR) (Feb. 2014).
`
`Page 6 of 16
`
`

`

`Docket No. FDA-2007—D—0369
`
`June 2, 2014
`Page 7
`
`(risperidone long acting injection) in August 2013.10 For these two products, FDA has
`recommended multiple-dose, steady-state, in vivo PK studies, with analysis of steady-state PK
`parameters Cmin ss, Cmax ss, and AUCimerdose, as well as percent fluctuation at steady state (as
`represented by the equation [100*(Cmax ss — Cmin ss)/Caverage ss]). For Risperdal Consta, FDA
`also recognized the need for an in vitro BE study in order to evaluate the product’s release
`properties. The revised guidance for Risperdal Consta therefore reflects FDA’s determination
`that statistical analysis of key time points in an in vitro release study, combined with a steady-
`state in vivo study, together provide the necessary basis for demonstrating BE.
`
`These developments represent renewed focus and consideration on the part of the Agency
`regarding the complex issues associated with demonstrating BE for long-acting depot products.
`They signal an informed recognition by the Agency of the need to assure adequate plasma
`exposure across the length of the dosing interval.
`
`RECOMMENDED CHANGES TO DRAFT GUIDANCE
`
`A.
`
`The AUC Metric AUC10_28 Should Be Added To The BE Analysis
`
`Vivitrol is a long—acting depot drug whose safety and efficacy is driven by a distinct,
`multiphasic release and absorption profile designed to maintain patients on therapeutic levels of
`naltrexone across the entire one month dosing interval. As detailed in our comments on the 2009
`Draft Guidance, Vivitrol’s single-dose PK profile is comprised of three clinically important
`phases. It has an initial phase during which naltrexone is immediately absorbed from the surface
`of the formulation’s microspheres into systemic circulation, followed by a hydration phase
`during which physical erosion of the microspheres begins, triggering the release of subsurface
`naltrexone leading to Cmax. This, in turn, is followed by the main sustained release phase, during
`which the microsphere formulation erodes in situ to allow a constant amount of drug to be
`absorbed, maintaining therapeutic plasma concentrations until the next scheduled monthly
`administration.
`
`A single intramuscular injection of Vivitrol will reside at the site of deposition for a
`_
`period of weeks, and the extent of naltrexone release at each of the three clinically important
`phases is a carefully controlled manufacturing and formulation—dependent variable.11 The timing
`
`10 Draft Guidance on Paliperidone Palmitate (Invega Sustenna) (Feb. 2014); Draft Guidance on Risperidone
`(Risperdal Consta) (Feb. 2014). Available at
`hm;://www.fda.gov/Drugs/GuidanceComplianceRegulatoglnformation/Guidances/ucm075207.htm.
`
`” For example, the release of excessive drug by a generic during the initial burst phase could be the result of a
`manufacturing process leading to a greater amount of unencapsulated naltrexone at the surface of the microsphere,
`and different solvents or other residues from the manufacturing process could cause injection site reactions resulting
`in local physiological changes that affect drug absorption such as a sharp drop in local pH. Moreover, Vivitrol
`shares the property of many extended release products in that absorption from the site of injection is slower than
`
`Page 7 of 16
`
`

`

`Docket No. FDA-2007-D-0369
`
`June 2, 2014
`
`Page 8
`
`and initiation of these phases is critical to the overall safety and effectiveness of the product. A
`proposed generic version of Vivitrol that exhibits a different release rate would interrupt the
`product’s carefully designed dosing intervals and may deleteriously interfere with clinically
`significant plasma-drug concentrations.
`
`Accordingly, as recognized by the Agency in the 2014 Draft Guidance, a combination of
`
`in vitro and in vivo BE studies are required in order to adequately evaluate the therapeutic
`equivalence of test and reference products. FDA is therefore correct to require a demonstration
`of equivalence in an in vitro study, covering the different release phases of the product.
`However, we believe the rate and extent of adsorption of Vivitrol cannot be adequately measured
`using an in vitro dissolution methodology alone, because of the complex physiological
`interactions between the product and surrounding tissues at the site of deposition, and the
`possibility of manufacturer—dependent differences. Nor can it be accounted for with FDA’s
`
`traditional BE parameters of Cmax and AUC. The in vivo study must be able to independently
`establish the rate and extent of absorption. Thus, in addition to an appropriately robust in vitro
`study, the Agency should recommend the use of additional BE parameters, as discussed below,
`
`consistent with its recent recommendations for similar products.
`
`Specifically, Alkermes developed an in silico population PK model for Vivitrol, allowing
`
`the Company to computationally simulate its PK profile and evaluate the robustness of various
`PK metrics in the context of a BB analysis.12 The validated model was used to conduct clinical
`trial simulations of in vivo BE studies of Vivitrol 380 mg against alternative naltrexone
`
`formulations designed to be either BE or non-BE to Vivitrol 380 mg. These simulations
`
`produced PK profiles of the reference drug and the comparator drug, enabling standard BE
`
`assessment based on measurement of particular PK metrics. That is, we found the averages of
`
`particular PK metrics for the subjects in the comparator and reference arms of the simulated PK
`studies, calculated their ratio, and then determined if the 90% confidence interval fell within the
`BE limit of 80-125%.13
`
`of the plasma curve is much more dependent on formulation than is the curve for an immediate release product
`where actual rate of elimination plays a larger part. See generally Yafiez, J.A., et al., Flip-flop pharmacokinetics —
`delivering a reversal ofdisposition: challenges and opportunities during drug development, THER DELIV. (2011)
`2(5): 643-672 (attached at Tab 2).
`
`12 A population PK model is well-suited for an analysis of Vivitrol’s pharmacokinetics. Population PK models can
`identify correlations between patient characteristics and observed drug concentrations, which allow the models to
`account for and simulate inter-subject variability. Food and Drug Administration, Guidancefor Industry:
`Population Pharmacokinetics at 2 (Feb. 1999),
`ht_tp://www.fda.gov/downloads/ScienceResearch/SpecialTopics/WomensHealthResearch/UCM133184.pdf.
`Alkermes previously submitted the results of a separate population PK analysis (report #ALK21-011) in support of
`Vivitrol’s original marketing application. Vivitrol Clinical Pharmacology and Biopharmaceutics Review(s) at p.6-7
`
`of PDF (Mar. 31, 2005), ht_tp://www.accessdata.fda.gov/drugsatfda docs/nda/2006/0218975000 ClinPharmedf.
`
`‘3 Food and Drug Administration, Guidancefor Industry: Statistical Approaches to Establishing Bioequivalence at 2
`
`Page 8 of 16
`
`

`

`Docket No. FDA—2007-D-0369
`
`June 2, 2014
`Page 9
`
`By observing the frequency at which a particular PK metric was able to correctly confirm
`or reject a comparator drug as BE in our simulations, we were able to quantify the ability of that
`metric to serve as an indicator of whether a comparator drug is BE to Vivitrol. Using this
`framework, we investigated two issues:
`
`0 Whether a standard BE analysis of Vivitrol’s PK profile — based on measurement
`of Cmax, AUCO-inf, and AUCo_t(28 days) — consistently and accurately identified
`naltrexone formulations that are BE to Vivitrol, as well as rejected naltrexone
`formulations that are non—BE to Vivitrol; and
`
`0 Whether a partial AUC metric would be a more accurate predictor of
`bioequivalence than standard BE measures. Specifically, we investigated whether
`measurement of AUC0_3, AUC3_10, AUC10-28, or AUC14_28 could serve as a more
`
`accurate metric to recognize naltrexone formulations that are BE to Vivitrol, as
`
`well as to reject naltrexone formulations that are non-BE to Vivitrol.
`
`Detailed information on the development and validation of our Vivitrol population PK
`model may be found in the Appendix to these comments. Briefly, we developed our model using
`standard population PK methodology.” To obtain parameter values, we fit the model to PK data
`from ALK21-005, a single-center, double-blind, placebo-controlled Alkermes study that
`evaluated the single- and repeat—dose pharmacokinetics of Vivitrol in healthy subjects. To
`account for inter-subj ect variability, we included in the model a stochastic component based on
`an analysis of individual patient demographics and their associated PK profiles from ALK21-
`005. A visual depiction of our model’s structure and a table of its parameter values may be
`found in the Appendix.
`
`We evaluated each PK metric in two stages. First, to assess the power of a PK metric to
`identify formulations of naltrexone that are BE to Vivitrol, we ran clinical trial simulations of the
`
`reference product versus the reference product — that is, we ran clinical trial simulations of
`Vivitrol 380 mg against itself, and then assessed the ability of each PK metric to correctly
`identify the formulations as BE in spite of the inter-subj ect variability. As seen in Figure 1, we
`found that Cmax had the lowest power to identify Vivitrol 380 mg as BE with itself, which is
`likely due to the high inter-subj ect variability observed for Cmax in ALK21-005. The AUCO-inf
`
`and AUC04 parameters had the highest power to recognize Vivitrol 380 mg as BE with itself.
`The partial AUC metrics — AUC0_3, AUC3_10, AUC10-23, and AUC14_23 — had varying intermediate
`degrees of power between Cmax and AUCO-inf to recognize Vivitrol 380 mg as BE with itself, with
`AUC3_10 and AUC10.28 performing better than the other partial AUC metrics.
`
`M W. Byon et al., Establishing Best Practices and Guidance in Population Modeling: An Experience With an
`Internal Population Pharmacokinetic Analysis Guidance, 2 CPT: PHARMACOMETRICS & SYSTEMS PHARMACOLOGY
`
`Page 9 of 16
`
`

`

`Docket No. FDA-2007-D-0369
`
`June 2, 2014
`
`Page 10
`
`1.00—
`
`Ill
`
`m
`$0.75»
`,_
`E
`8 o 50
`< .
`2
`g.
`3
`g 025‘
`
`—
`
`E
`
`l
`25
`
`I
`|
`75
`50
`Number of Individuals Per Arm
`
`160
`
`M tr‘c
`8 '
`-e- CMAX
`a. AUCO_inf
`l- AUOD_28
`+ nuco_3
`@- Aucs_1o
`.
`“
`» AUCIO_28
`- AUCI4_28
`
`Figure 1: Probability of concluding bioequivalence of Vivitrol 380 mg (Reference) to
`Vivitrol 380 mg (Test) based on naltrexone pharmacokinetic parameters
`
`In the next stage of our analysis, we assessed the ability of each PK metric to reject two
`theoretical comparator formulations of naltrexone that were unlikely to be therapeutically
`equivalent to Vivitrol 380 mg. We framed the PK profile of these theoretical naltrexone
`formulations to emphasize potential differences between the 190 mg strength of Vivitrol, which
`was not efficacious in a clinical study of alcohol dependence, and the FDA approved 380 mg
`dose strength of Vivitrol. A significant number of patients receiving Alkermes’ investigational
`naltrexone 190 mg in ALK21-005 had naltrexone concentrations below 1 ng/mL during the
`sustained release phase. Notably, data from a clinical study wherein hydromorphone challenges
`were administered to test the ability of Vivitrol to block opioid agonism (ALK21-004)
`demonstrated that naltrexone concentrations above 1 ng/mL were sufficient to provide blockade.
`These data were utilized to substantiate investigation, and ultimately FDA approval, of Vivitrol
`for the treatment of opioid dependence. Thus, to simulate naltrexone formulations that were
`
`non-BE to Vivitrol, we designed the two following theoretical naltrexone formulations with
`diminished naltrexone concentrations —— i.e., less than 1 ng/mL — during Vivitrol’s sustained
`release phase:
`
`0 Naltrexone double release (NTX-DR) had a biphasic profile with a more
`pronounced initial peak at about 2 days, a secondary peak occurring at
`approximately 7 days, and a subsequent decline with no tertiary peak; and
`
`o Naltrexone single release (NTX-SR) had a single primary peak at about 2 days.
`
`Profile comparisons of NTX-DR 380 mg and NTX-SR 380 mg to Vivitrol 380 mg are
`shown in Figures 2 and 3, respectively, with the dashed line indicating a naltrexone
`
`Page 10 of 16
`
`

`

`Docket No. FDA—2007-D-0369
`
`June 2, 2014
`Page 11
`
`naltrexone formulations have prolonged periods with naltrexone concentrations less than 1
`ng/mL towards the tail ends of their PK profiles. Thus, like Alkermes’ naltrexone 190 mg, we
`would expect the simulated comparator naltrexone formulations to show a lack of efficacy and
`thus be non-BE to Vivitrol 380 mg. In this context, a strong PK metric would have a high
`probability of rejecting the comparator naltrexone formulations as non-BE to Vivitrol 380 mg.
`
`formulation
`‘I-Mviuol
`1,.
`
`E NTX—DR
`
` 0
`
`1O
`
`20
`TIME (days)
`
`1?
`
`i Z
`
`g
`n
`
`E
`
`0 8oO
`
`Figure 2: Population mean derived naltrexone concentration following administration of
`
`Vivitrol 380 mg (Reference) and NTX-DR (Comparator)
`
`formulation
`.,
`,I-IVuvinoI
`:
`-. "TX—SR
`
`
`
`15
`
`3 i:
`
`10
`a
`:9
`h
`E
`fl)
`
`8 5o
`o
`
`O
`
`O
`
`10
`
`20
`um: (days)
`
`30
`
`40
`
`Figure 3: Population mean derived naltrexone concentration following administration of
`
`Vivitrol 380 mg (Reference) and NTX—SR (Comparator)
`
`As seen in Figures 4 and 5, our simulations of Vivitrol versus NTX-DR and NTX-SR,
`
`respectively, found that the standard BE metrics performed poorly in distinguishing Vivitrol
`
`Page 11 of 16
`
`

`

`Docket No. FDA-2007-D-0369
`
`June 2, 2014
`Page 12
`
`greater than 15 individuals per arm. This raises the possibility that FDA’s recommended in vivo
`BE test, which relies on measurement of the standard BE metrics, is prone to type 1 error — i.e.,
`FDA’s test carries the risk that it will accept an alternative naltrexone formulation as BE to
`Vivitrol even if the alternative formulation exhibits fundamentally different release properties
`and therapeutic deficit (subtherapeutic PK) in the latter part of the dosing interval. On the other
`hand, our simulations of Vivitrol versus NTX-DR and NTX-SR found that partial AUC metrics
`performed strongly in distinguishing Vivitrol from the comparator naltrexone formulations.
`Three of the partial AUC metrics we tested — AUC3_10, AUC10_23, and AUC14_28 — rejected NTX-
`DR and NTX-SR as BE to Vivitrol across our simulations regardless of the number of
`individuals per arm.15
`
`1.00 —
`
`NI
`
`n a
`
`8 0'75_
`o
`g.
`.3
`gaso—
`2
`E
`€025"
`E
`
`0.00 —
`
`.
`Metric
`ae-CMAx
`*AUODJnf
`l-Auco_2a
`+Auco_3
`E- AU03_10
`»
`1 mamas
`u-AUC‘I4_28
`
`25
`
`7'5
`5‘0
`Number of Individuals Per Arm
`
`100
`
`Figure 4: Probability of falsely concluding that NTX-DR 380 mg (Comparator) is
`
`bioequivalent to Vivitrol 380 mg (Reference) using traditional and proposed
`pharmacokinetic parameters
`
`15 As a general rule, increased clinical trial sample size results in an increased probability — due to reduction in
`statistical variability — that BE criteria will be met. But as seen in figures 4 and 5, as more individuals were added to
`each study arm in clinical trial simulations of Vivitrol, there was a greater likelihood that certain BE metrics would
`falsely accept a non-bioequivalent product as bioequivalent. Specifically, as the number of individuals per arm
`increased, we observed a corresponding increase in probability that PK metrics that were not sensitive to naltrexone
`concentration changes relative to the therapeutic threshold of 1 ng/mL (such as AUCOM) would result in a false
`acceptance of comparator formulations as BE to the reference formulation. However, for BB metrics that were
`sensitive to naltrexone concentration changes relative to the therapeutic threshold of 1 ng/mL (such as AUC10.28), the
`probability of falsely accepting the comparator formulations as BE did not increase regardless of the number of
`
`Page 12 of 16
`
`

`

`Docket No. FDA-2007-D-0369
`
`June 2, 20 14
`Page 13
`
`1.00 —
`
`III
`in
`a
`8 015 _
`2
`a
`fl
`E 0.50—
`'3.
`E
`3 0‘25 *
`
`32n
`
`.
`
`0.00 —
`
`.
`Metric
`li- CMAX
`‘AUCOJnf
`I-Auco_2a
`+ AU00_3
`a-AU03_10
`; AUC! 0.28
`AUC14_28
`
`Number of Individuals Per Arm
`
`Figure 5: Probability of falsely concluding that NTX-SR 380 mg (Comparator) is
`bioequivalent to Vivitrol 380 mg (Reference) using traditional and proposed
`pharmacokinetic parameters
`
`As noted previously, AUC3_10 and AUC10.23 performed better than the other partial AUC
`metrics in recognizing Vivitrol 380 mg as BE with itself (Figure 1). When combined with their
`superior ability to distinguish Vivitrol from the comparator naltrexone formulations, AUCHO and
`AUC10_23 emerge as compelling candidates for an additional metric to measure as a necessary
`supplement to FDA’S existing standards to ensure BE.
`
`In particular, analysis of AUC10_28 for BE provides a statistically robust test of the
`sustained release phase of Vivitrol’s PK profile. Furthermore, this segment of the PK profile
`encompasses the part of the curve that we believe may be most critical to assuring therapeutic
`equivalence of a generic sustained release inj ectable formulation of naltrexone.
`
`As discussed above, Vivitrol is intended to maintain plasma levels at therapeutic levels
`over a one month dosing interval. It achieves this through its multiphasic release profile and
`sustained release properties. Specifically, the formulation allows a single 380 mg dose of
`naltrexone to release an adequate amount of drug almost immediately and to maintain plasma
`concentrations at a clinically effective level for the entire dosing interval. For example, our
`studies of naltrexone blockade of the hydromorphone opioid response showed that a single dose
`of 300 mg of Vivitrol can maintain plasma concentrations at or above 1 ng/mL at day 28.16 In
`contrast, subjects receiving a 150 mg dose of the same formulation showed plasma levels below
`1 ng/mL at day 28 in all subj ects.17 These data help to illustrate that effective naltrexone
`
`'6 Vivitrol Clinical Pharmacology and Biopharmaceutics Review(s) at p.14-17 ofPDF (Mar. 31, 2005),
`ht_tp://www.accessdata.fda.gov/drugsatfda docs/nda/2006/0218975000 C1inPharmR.pdf.
`
`Page 13 of 16
`
`

`

`Docket No. FDA-2007—D-0369
`
`June 2, 2014
`
`Page 14
`
`blockade is a function of plasma concentration, and that a dose above 300 mg is required in order
`to maintain plasma levels above 1 ng/mL at the end of the dosing interval.
`
`Further, in clinical studies comparing 380 mg and 190 mg versus placebo, the 190 mg
`dose failed to show effectiveness.18 The PK profile differences between the 380 mg and
`investigational 190