`RESEARCH
`
`
`
`APPLICATION NUMBER:
`
`207620Orig1s000
`
`OFFICE DIRECTOR MEMO
`
`
`
`
`
`
`
`
`
`
`
`Date
`From
`
`Office of Drug Evaluation-I: Decisional Memo
`
`
`July 7, 2015
`Ellis F. Unger, MD, Director
`Office of Drug Evaluation-I, Office of New Drugs, CDER
`Office Director Decisional Memo
`207620
`Novartis Pharmaceuticals Corporation
`December 17, 2014
`August 17, 2015
`Entresto
`sacubitril and valsartan
`24 mg sacubitril and 26 mg valsartan; film-coated tablets
`49 mg sacubitril and 51 mg valsartan; film-coated tablets
`97 mg sacubitril and 103 mg valsartan; film-coated tablets
`
`Subject
`New Drug Application (NDA) #
`Applicant Name
`Date of Submission
`PDUFA Goal Date
`Proprietary Name/
`Established (USAN) Name
`Dosage Forms/ Strengths
`
`Indication originally sought by
`applicant (see page 29 for final)
`
`Approval
`
`Medical Officer Clinical Review
`
`Action:
`
`Material Reviewed/Consulted - Action Package, including:
`Project Manager
`Alexis Childers, RAC
`Kimberly Smith, MD (Efficacy), Tzu-Yun McDowell, PhD
`(Safety)
`Luning Zhuang PhD, Sreedharan Sabarinath, PhD
`Clinical Pharmacology/Pharmacometrics
`John Lawrence, PhD; James Hung, PhD
`Statistical Review
`William Link, PhD, Al De Felice, PhD
`Pharmacology Toxicology
`Executive Cancer Assessment Committee Paul Brown, PhD (acting chair)
`Wendy Wilson-Lee, PhD (technical lead), Anamitro
`Banerjee, PhD (drug substance), Sherita McLamore-
`Hines, PhD (drug product), Bogdan Kurtyka, PhD
`(process), Zhong Li, PhD (facility)
`Sharon K. Gershon, PharmD
`Miriam Dinatale, DO
`Salaheldin Hamed, PhD
`Mohammad Rahman, PhD; Karl Lin, PhD
`
`Office of New Drug Quality Assessment
`
`Office of Scientific Investigations
`Division of Pediatric and Maternal Health
`Biopharmaceutics Review
`Carcinogenicity Study
`Division of Medication Error Prevention and
`Analysis
`Risk Management Review
`Cross-Discipline Team Leader
`Director, Division of Cardiovascular and
`Renal Products
`
`
`
`Janine Stewart, PharmD; Alice Tu, PharmD
`
`Danny Gonzales, PharmD, MS, Kim Lehrfeld, PharmD
`Aliza Thompson, MD
`
`Norman Stockbridge, MD, PhD
`
`Reference ID: 3788923
`
`Office Director Decisional Memo – NDA 207620 – Page 1 of 29
`
`(b) (4)
`
`
`
`1 .
`
`Introduction
`
`Novartis is seeking approval of LCZ696 for the proposed indication:
`
`m4)
`
`With a number of changes to the label, including changes to the indication statement (see
`Summary/Conclusions), the review team endorses approval, and I agree with their
`recommendation.
`
`2.
`
`Background
`
`Description:
`
`M“), a type of sodium salt complex, consisting of valsartan and sacubitril
`LCZGQG is a
`anions, sodium cations, and water. These 4 individual components are present in a 22:65
`molar ratio, and are not ionically bound. The drug product contains the LCZ696
`(mm as
`the active ingredient. The active moieties in the LC2696
`(him are sacubitril and valsartan.
`
`The description of the chemical nature of the active ingredient was an important issue that had
`to be negotiated with the applicant. The Office of Pharmaceutical Quality (OPQ) initially
`recommended use of the term
`"W“ to describe the tablet’s active ingredient in Section 11
`of labeling, whereas the applicant had proposed the term
`"’“4’
`
`In an addendum to the Quality review, OPQ noted that both descriptions correctly represent the
`chemical nature of the active ingredient and are scientifically valid. The structural X-ray
`diffraction data submitted demonstrate that the active ingredient meets the criteria delineated in
`FDA Guidance
`W"
`
`The active ingredient
`can also be considered a complex, however, based on the IUPAC Gold Book definition: a
`molecular entity formed by loose association involving two or more molecular entities (ionic or
`neutral); bonding is normally non-covalent.
`
`mm the applicant’s preferred description of the active ingredient,
`correct, OPQ noted that the term
`OPQ is recommending use of the term “complex" be used to refer to the
`active ingredient in Section 11 of labeling, and the applicant has agreed.
`
`(m4) is
`“M"
`
`Valsartan is a previously approved molecular entity, an angiotensin II receptor blocker (ARB),
`which is widely marketed for hypertension and heart failure as Diovan and generics. Sacubitril
`is a neprilysin inhibitor, a first-in-class new molecular entity (NME), although there is some
`experience with this class of agents, as discussed later in this memo.
`
`Reference ID: 3788923
`
`Office Director Decisional Memo — NDA 207620 — Page 2 of 29
`
`
`
`W", it dissociates in vivo to the active
`Although the active ingredient in the tablet is a
`moieties valsartan and sacubitril, and so it has been consistently viewed as a combination
`product from a regulatory perspective.
`
`Disease Background:
`
`Over 5 million people in the US have heart failure, about half of whom have reduced left
`ventricular ejection fraction or systolic heart failure. (Many patients with heart failure have
`preserved left ventricular systolic function, so-called “diastolic heart failure,” for which there are
`no approved treatments.) According to the 2013 American College of Cardiology
`Foundation/American Heart Association “Guideline for the Management of Heart Failure," the
`lifetime risk of developing heart failure is 20% for the US. population 2 40 years of age, with
`over 650,000 new cases diagnosed annually (J Am Coll Cardiol 2013;e147—239). The incidence
`of heart failure increases with age: from ~2 per 100 individuals at age 65 to 69 to over 8 per 100
`individuals at age 85 and over. As life expectancy increases in the US, the prevalence is
`anticipated to rise. Moreover, despite improvements in the pharrnacologic and non-
`pharmacologic management of heart failure, 5-year survival rates are still only ~50%.
`
`There is an excellent summary of current therapy for heart failure in the Clinical Review, page
`14.
`
`For the regulatory history, refer to the clinical review and the cross—discipline team leader
`review.
`
`3.
`
`Product Quality
`
`OPQ recommends approval from a drug product perspective.
`
`“M" comprised of two active moieties
`As noted above, the active ingredient in LCZ696 is a
`- sacubitril and valsartan anions — with 1:1 stoichiometry. The other components are sodium
`cations and water. The
`M" is synthesized using the drug substances sacubitril
`“W”
`w, p,
`(I!) (‘9
`and valsartan
`
`dissociate in ViVO to release sacubitril and valsartan.
`
`. The
`
`W" quickly
`
`Although designated as regulatory drug substances, it was agreed that the applicant's quality
`systems and standards control sacubitril
`“m" and valsartan
`“M"
`to produce the oo-crystal.
`(m4) specifications for(mu)
`
`sacubitril
`
`(m4) and valsartan
`
`include appropriate tests and acceptance criteria to ensure the identity, purity,
`strength, quality, and bioavailability of these compounds.
`
`A 24-month drug product expiration date has been granted when stored at room temperature
`and protected from moisture in the intended container closure. The drug product is packaged in
`bottles and unit dose blister packages.
`
`Based on firm inspectional history and data reviewed during the pre—approval inspections, OPQ
`found the manufacturing facilities to be acceptable.
`
`Reference ID: 3788923
`
`Office Director Decisional Memo — NDA 207620 — Page 3 of 29
`
`
`
`Following discussions regarding the expression of strength in the carton and container labels,
`OPQ agreed on a compromise to allow use of a “I” between the sacubitril and valsartan in the
`established name, and found the carton and container labels acceptable.
`
`Post-Marketing Commitment:
`
`It was determined during the review that the dissolution data submitted for the clinical and the
`registration batches of the "’""mg strength did not support the dissolution acceptance criterion
`proposed by the applicant (Q = 93% at {2} minutes).
`M“)
`
`OPQ recommended that the applicant optimize the dissolution method and acceptance criterion,
`which would require a post-marketing commitment.
`It was noted that the control strategy for the
`current product (e.g., operating closely to the normal operating ranges for the clinical trial batch)
`would ensure the quality of the drug product.
`
`In the absence of an adequate in vitro to in vivo relation and proper exposure-response data,
`FDA recommended establishment of a release specification at Q = g;% to ensure complete
`release of the drug substance.
`
`There will be a post-marketing commitment with the following goals: 1) Development of a new
`dissolution method for all the strengths with demonstrated discriminating ability,
`mm
`
`and 2) Setting of the final dissolution acceptance criterion for Entresto
`(sacubitril/valsartan) Tablets, 97/103, 49/51, and 24/26 mg using the new method and the
`overall multipoint dissolution profile data from a minimum of 12 commercial batches/strength,
`manufactured under the same conditions as those used for the manufactured of the batches
`
`used in pivotal clinical trials.
`
`The FDA would be open to the possibility of a
`
`W"
`
`OPQ reiterated that a justification would need to be provided, supported by data,
`before agreeing to
`(”m’ for the drug product.
`
`The applicant has agreed to this post-marketing commitment, and agreed to submit a
`development report by February 1, 2016, and the final report by July 1, 2016.
`
`4.
`
`Nonclinical Pharmacology/Toxicology:
`
`Salient findings in mice, rats, rabbits, and cynomolgus monkeys included renal juxtaglomerular
`hypertrophy/hyperplasia, renal tubular changes; decreased hemoglobin/hematocrit and
`reticulocytes; decreased heart weights (without histopathological findings); reversible focal
`gastric mucosal erosion, and emesis and diarrhea without histologic correlates. According to
`Dr. Link, these findings do not raise concerns for human use because they reflect adaptive
`responses and/or exaggerated phannacodynamic responses to high doses.
`
`Reference ID: 3788923
`
`Office Director Decisional Memo — NDA 207620 — Page 4 of 29
`
`
`
`
`
`
`
`Genotoxicity assays of LCZ696, sacubitril and LBQ657, sacubitril’s active metabolite, were
`negative.
`
`LCZ696 had no effect on fertility in rats. Both sacubitril and valsartan are known to cause fetal
`toxicity and embryo-fetal lethality in rabbits, and LCZ696 increased embryo lethality in both rats
`and rabbits. LCZ696 was teratogenic in rabbits at ≥10 mg/kg. As for all drugs that act directly
`on the renin-angiotensin-aldosterone system, LCZ696 will be contraindicated during pregnancy.
`
`Neprilysin is a major beta amyloid-degrading enzyme in the brain. There is, therefore, a
`theoretical risk that LCZ696, by inhibiting neprilysin, could cause accumulation of β-amyloid
`(Aβ) in the brain, leading to cognitive impairment. There was much interest, therefore, in the
`non-clinical studies designed to examine LCZ696’s effect on Aβ.
`
`The applicant assessed the effects of LCZ696 on Aβ concentrations in brain and cerebrospinal
`fluid (CSF) in young (2.5 to 4 year-old) female cynomolgus monkeys treated with LCZ696, 50
`mg/kg/day, for 2 weeks. Treatment was associated with increases in Aβ1-40, Aβ1-42, and Aβ1-38 in
`cerebrospinal fluid (CSF), without corresponding increases in brain. The Division of Neurology
`Products (DNP) was consulted, in part to review this study. DNP noted that sufficient levels of
`LBQ657 reached the central nervous system (CNS) to inhibit neprilysin, but that other Aβ
`clearance mechanisms, including transport into the CSF, compensated, such that there was no
`apparent net increase in brain Aβ at steady state.
`
`DNP also had two major criticisms of the study. First, DNP stated that the LCZ696 dose used
`should have been higher. The applicant deemed the LCZ696 dose tested to be “clinically
`relevant,” based on similar concentrations of LBQ657 achieved in the CSF of monkeys and
`healthy volunteers given the maximum recommended human dose (MRHD) – a comparison of
`exposures based on Cmax. Based on AUC0-24 hr, however, DNP noted that exposure in monkeys
`was approximately half the exposure in humans. DNP would have preferred testing of higher
`doses, to provide CNS exposures several-fold higher than those expected at the MRHD.
`Second, DNP expressed the view that effects observed in young monkeys may not be
`predictive of effects in elderly humans, because cynomolgus monkeys do not typically develop
`measurable cerebral amyloid pathology until they are much older.
`
`Study 0670621 was a 39-week toxicology study where 12 young (2 to 4 year-old) cynomolgus
`monkeys received LCZ696 300 mg/kg/day (AUC exposure ~2X the MRHD), and there was no
`Aβ42 immunostaining in the brain (tissues with known Aβ positivity were included to establish
`assay sensitivity). DNP also questioned the informativeness of this study, however, because of
`the young age of the monkeys. DNP suggested a study in aged monkeys, to quantify levels of
`soluble and insoluble Aβ in brain homogenates, and to assess immunoreactive Aβ.
`
`Carcinogenicity:
`
`Two-year carcinogenicity studies of sacubitril were performed in rats and mice, and there were
`no effects on tumor incidence or survival. The Executive Carcinogenicity Assessment
`Committee found the studies acceptable.
`
`5.
`
`
`Clinical Pharmacology
`
`Reference ID: 3788923
`
`Office Director Decisional Memo – NDA 207620 – Page 5 of 29
`
`
`
`
`
`
`
`The Clinical Pharmacology team recommends approval, with a lower starting dose in patients
`with severe renal function impairment and moderate hepatic impairment. The applicant has
`accepted these recommendations.
`
`Mechanism of action: LBQ657, the active metabolite of sacubitril, inhibits neprilysin, which
`causes proteolytic degradation and inactivation of natriuretic peptides, including atrial natriuretic
`peptide (ANP), brain natriuretic peptide (BNP), and C-Type natriuretic peptide (CNP). These
`peptides activate membrane-bound guanylyl cyclase-coupled receptors and increase
`concentrations of cyclic guanosine monophosphate (cGMP). With inhibition of the inactivation
`of these proteins, LBQ657 is thought to promote vasodilation, natriuresis and diuresis, increase
`renal blood flow and glomerular filtration, inhibit renin and aldosterone release, and reduce
`sympathetic activity. By blocking the binding of angiotensin II to the AT1 receptor, valsartan
`blocks the vasoconstrictor and aldosterone-secreting effects of angiotensin II, cardiac
`stimulation, and renal reabsorption of sodium.
`
`Pharmacokinetics:
`
` •
`
`•
`
` Upon oral administration, LCZ696 dissociates into sacubitril and valsartan; both moieties are
`rapidly absorbed without significant food effects.
`• Sacubitril is a pro-drug that undergoes metabolism via esterases to form LBQ657, which
`inhibits neprilysin. LBQ567 is not further metabolized. Valsartan does not undergo
`significant metabolism.
`• Absolute bioavailability of sacubitril from LCZ696 is ≥ 60%. The bioavailability of valsartan
`from LCZ696 is at least 50% higher than valsartan when valsartan is administered alone.
`For example, valsartan from 400 mg LCZ696 (containing ~ 203 mg valsartan) is
`approximately equivalent to 320 mg of the marketed valsartan formulation.
`• Sacubitril and valsartan are highly protein-bound (97% and 94%, respectively).
`• There is no significant CYP isozyme involvement in the metabolism of sacubitril or valsartan.
`• Approximately 52-68% of sacubitril is excreted in urine (as LBQ657) and 37-48% recovered
`in feces. Approximately 83% of valsartan is excreted in feces and about 13% in urine.
`In healthy subjects, the average elimination half-lives of sacubitril, LBQ657, and valsartan
`are 1.4, 11.5 and 9.9 hours, respectively.
`• Age, sex, race, and weight have little effect on exposure.
`• Renal impairment: Steady state exposure of LBQ657 increases by about 2-fold in patients
`with all degrees of renal impairment (mild to severe), whereas effects on valsartan exposure
`were minimal. No study was conducted in patients on dialysis, but LBQ657 and valsartan
`are highly protein-bound and unlikely to be removed by dialysis. Note: out of caution, the
`recommended starting dose will be halved in patients with severe renal impairment (eGFR <
`30 mL/min/1.73 m2).
`• Hepatic impairment: In patients with mild hepatic impairment, exposures of sacubitril,
`LBQ657 and valsartan are increased only slightly relative to healthy subjects. In patients
`with moderate hepatic impairment, exposures of sacubitril, LBQ657, and valsartan were
`increased by ~245%, 90%, and 109%, respectively, relative to healthy subjects, and the
`recommended starting dose will be halved in these patients. No studies were conducted in
`patients with severe hepatic impairment, where use of the drug will not be recommended.
`
`
`Pharmacodynamics:
`
`
`Reference ID: 3788923
`
`Office Director Decisional Memo – NDA 207620 – Page 6 of 29
`
`
`
`Clinical Microbiology
`
`Clinical/Statistical Efficacy
`
`
`
`
`
`Aβ concentrations: In 39 healthy subjects, administration of LCZ696 400 mg once a day for 2
`weeks was associated with a 42% increase in CSF Aβ1-38 relative to baseline and a 50%
`increase in plasma Aβ1-40. As explained in the clinical review, the clinical significance of these
`findings is unknown.
`
`QT Effects:
`
`QT effects: No significant QTc prolongation was observed with LCZ696 (400 mg and 1200 mg)
`in a thorough QT study.
`
`6.
`
`The drug is not an antimicrobial. According to the product quality review, the tests and
`proposed acceptance criteria for microbial burden are adequate.
`
`7.
`
`The evidence of efficacy for LCZ696 is provided by PARADIGM-HF, described in detail below.
`The results of this study were published recently (McMurray JJ, et al: Angiotensin–neprilysin
`inhibition versus enalapril in heart failure. N Engl J Med 2014; 371:993).
`
`PARADIGM-HF
`
`Novartis conducted PARADIGM-HF, a single phase 3 outcome trial, in support of the proposed
`indication. PARADIGM-HF was an international, randomized, double-blind, double-dummy,
`event-driven, active-control trial comparing LCZ696 with enalapril in adult patients with NYHA
`class II to IV chronic heart failure and left ventricular ejection fraction (LVEF) ≤ 40% (changed to
`≤ 35% per protocol amendment 1), who were able to tolerate both of the test drugs during run-in
`phases (i.e., the study was enriched for patients who could tolerate the drugs).
`
`Although the population has been described as having “stable” heart failure, the meaning of
`“stable” is not straightforward here and is somewhat of a misnomer. Patients were to have been
`on stable doses of a beta-blocker (unless contraindicated or poorly tolerated) and an ACE
`inhibitor or an ARB for ≥ 4 weeks prior to screening. On the other hand, patients had to have B-
`type natriuretic peptide (BNP) ≥ 150 pg/mL, or ≥ 100 pg/mL with a hospitalization for heart
`failure within the last 12 months (an enrichment maneuver).
`
`PARADIGM-HF was essentially an add-on study that tested the concept that LCZ696, a
`combination of sacubitril and a RAAS inhibitor (valsartan), was superior to a RAAS inhibitor
`alone (enalapril).
`
`The 1° endpoint was a composite endpoint of time-to-first heart failure hospitalization or
`cardiovascular (CV) death. Secondary endpoints included:
`
` •
`
`time to all-cause death;
`
`• change from baseline to Month 8 in the clinical summary score for heart failure symptoms
`and physical limitations as assessed by the Kansas City Cardiomyopathy Questionnaire
`(KCCQ);
`time to new-onset of atrial fibrillation;
`
`•
`
`Reference ID: 3788923
`
`Office Director Decisional Memo – NDA 207620 – Page 7 of 29
`
`
`
`time to first occurrence of: a 50% decline in estimated glomerular filtration rate (eGFR)
`relative to baseline; a > 30 mL/min/1.73 m2 decrease in eGFR to a value < 60 mL/min/1.73
`m2; or end-stage renal disease (ESRD).
`
`
`
`
`
`•
`
` A
`
` Clinical Endpoint Committee adjudicated all reported deaths, unplanned hospitalizations for
`heart failure and myocardial ischemia, non-fatal myocardial infarctions, non-fatal strokes,
`resuscitated sudden deaths, new-onset atrial fibrillation, new-onset diabetes mellitus, ESRD,
`and worsening renal function events that occurred during the run-in and randomized periods.
`
`After discontinuing their ACE inhibitor or ARB, patients entered sequential single-blind run-in
`periods during which they received enalapril 10 mg BID, followed by LCZ696 100 mg BID,
`increasing to 200 mg BID. Subjects who completed the sequential run-in periods were
`randomized 1:1 to LCZ696 200 BID or enalapril 10 mg BID (see Figure 1).
`
`
`Figure 1: PARADIGM-HF Study Design
`
`
`
`
`Based on various assumptions, 2,410 composite endpoint events were expected to provide
`97% power to detect a 15% reduction in the 1° endpoint.
`
`Three (3) interim analyses were planned to assess efficacy: when 1/3, 1/2, and 2/3 of the
`expected 1° endpoint events were reported. A Haybittle-Peto boundary rule was used to assess
`superiority and control the overall Type-I error at 0.025 (1-sided). An α of 0.0001 (one-sided)
`was spent at the first interim analysis; with 0.001 (one-sided) spent at the second and third
`interim analyses. If the study were stopped at an interim analysis, the 2° endpoints were to be
`tested using the same α used for the 1° endpoint. The testing of the 4 secondary endpoints
`was to be done using the Bonferroni-Holm's method. As shown in Figure 1 of Dr. Lawrence’s
`review, the α was initially split between the first two secondary endpoints, with 80% of the α
`allocated to all-cause mortality and 20% allocated to the KCCQ. If both hypotheses were
`rejected, the full α was to be allocated to the next secondary endpoint on the testing chain; if
`only one of the initial hypotheses was rejected, the α allocated to the rejected hypothesis would
`then be allocated to the next 2° endpoint.
`
`
`Reference ID: 3788923
`
`Office Director Decisional Memo – NDA 207620 – Page 8 of 29
`
`
`
`
`
`
`
`According to the statistical analysis plan, the 1° composite endpoint would be analyzed using a
`Cox regression model with terms for treatment and region. The 1° efficacy analysis was to
`include all positively adjudicated events occurring between randomization and March 31, 2014
`(the date the trial was terminated early for efficacy). The analysis was to be based on all
`randomized patients, excluding patients who did not qualify for randomization but were
`inadvertently randomized and did not receive study drug.
`
`There were no controversies or disagreements with the applicant with respect to the statistical
`plan or analyses.
`
`Results:
`
`Disposition:
`The trial was initiated on December 8, 2009 and terminated for efficacy on March 31, 2014 on
`the basis of the third planned interim analysis. A total of 10,521 subjects entered the initial run-
`in period. Of these, 1,102 subjects failed the enalapril run-in period and 982 failed the LCZ696
`run-in period (10.5% and 10.4% of subjects entering each run-in period, respectively). In total,
`approximately 20% of those who entered the initial run-in period were not randomized. About
`half of these patients discontinued because of an adverse event, most commonly renal
`dysfunction, hyperkalemia, or hypotension. Importantly, as the review team points out, because
`of the run-in periods, the randomized population was enriched in terms of tolerating the drug
`and their willingness to stay in the trial. In “real-world” use, of course, renal dysfunction,
`hyperkalemia, and hypotension are expected more often than reported in the randomized phase
`of PARADIGM-HF.
`
` total of 8,442 subjects were randomized, and all but 10 received study drug. A total of 35
`(0.4%) subjects did not complete study follow-up because of withdrawal of consent or loss to
`follow-up. In the double-blind treatment period, ~17% of subjects prematurely discontinued
`therapy in the LCZ696 group, compared with ~19% in the enalapril group. The most common
`reason for treatment discontinuation during the double-blind treatment period was an adverse
`event (10.4% of subjects randomized to LCZ696 and 12.1% of subjects randomized to
`enalapril). Vital status was unknown for 20 (0.2%) subjects.
`
`Serious GCP violations were identified at 4 sites that had enrolled a total of 37 subjects. The
`applicant prospectively chose to exclude all 37 from the efficacy analyses, but included them in
`the safety analyses. Six (6) subjects were “misrandomized:” IVRS randomization calls were
`mistakenly placed, in spite of the subjects failing the run-in period. None of the 6 received study
`medication, and all were prospectively excluded from efficacy analyses.
`
`
` A
`
`Reference ID: 3788923
`
`Office Director Decisional Memo – NDA 207620 – Page 9 of 29
`
`
`
`
`
`
`
`
`
`
`Table 1: PARADIGM-HF - Subject disposition during the randomized double-blind period
`
`Enalapril
`LCZ696
`n (%)
`n (%)
`4233 (100)
`4209 (100)
`4 (0.1)
`6 (0.1)
`4212 (99.5)
`4187 (99.5)
`21 (0.5)
`22 (0.5)
`2 (0.1)
`4 (0.1)
`18 (0.4)
`19 (0.5)
`3441 (81.8)
`3379 (79.8)
`2869 (67.8)
`3011 (71.5)
`815 (19.3)
`729 (17.3)
`18 (0.4)
`17 (0.4)
`13 (0.3)
`15 (0.4)
`4 (0.1)
`9 (0.2)
`5 (0.1)
`2 (0.1)
`
`Randomized
` Not treated
`Primary efficacy population (full analysis set)
` Excluded
` Misrandomized1
` Site excluded for GCP violations
`Completed study on treatment
` Alive at study termination
`Prematurely discontinued study treatment
`Did not complete study
` Withdrew consent
` Vital status unknown
` Lost to follow-up (vital status unknown)
`
`
`As expected for an 8,000-patient study, the two treatment arms were balanced with respect to
`baseline characteristics. Approximately 5% of subjects were enrolled at U.S. sites. Mean age
`was 64 years, with ~19% of patients over 75. The majority of subjects were Caucasian (66%)
`and male (78%). Approximately 5% of subjects were black. Most subjects (70%) were NYHA
`Class II; 24% were Class III and 0.7% were Class IV. Mean ejection fraction was 29%; mean
`baseline eGFR was 68 mL/min/1.73m2; and mean systolic blood pressure and heart rate were
`122 and 73, respectively. The majority of subjects were taking beta-blockers (94%),
`mineralocorticoid antagonists (56%), and diuretics (82%). Most patients (71%) had a history of
`hypertension. Atrial fibrillation was reported in 37% of patients. The cause of heart failure was
`reported to be ischemic in 60% of patients.
`
`Patients were reasonably representative of a U.S. heart failure population with respect to
`demographics (race notwithstanding) and disease-specific factors. But patients were not
`representative with respect to use of implantable cardioverter-defibrillators (ICD) or cardiac
`resynchronization therapy-defibrillator (CRT-D). Only 15% of the overall study population used
`these devices; in contrast, device use was reported in 60% of subjects enrolled at U.S. sites.
`Thus, relative to the U.S. population, patients with ICDs or CRT-D were under-represented in
`the study. The Registry to Improve the Use of Evidence-Based Heart Failure Therapies in the
`Outpatient Setting included ~35,000 patients at 167 US outpatient cardiology practices with
`reduced LVEF (≤ 35%) and chronic HF or previous myocardial infarction. Based on chart
`review, use of a CRT with a pacemaker or defibrillator ranged from 37 to 66%; use of ICDs
`ranged from 50 to 77% (Circulation 2010;122:585).
`
`Median duration of exposure was ~2 years in both treatment groups.
`
`Primary endpoint findings:
`
`As of the interim analysis cut-off date, there were 914 endpoint events in the LZ696 arm
`(21.8%) and 1117 events (26.5%) in the enalapril arm (HR = 0.80; 95% confidence interval [CI]
`0.73; 0.87, p < 0.001). See Table 2 and Figure 2. The treatment effect reflected reductions in
`both of the endpoint components. Approximately 40% of first events were cardiovascular
`deaths; ~60% were hospitalizations for worsening heart failure.
`
`Reference ID: 3788923
`
`Office Director Decisional Memo – NDA 207620 – Page 10 of 29
`
`
`
`
`
`
`
`
`Table 2: PARADIGM-HF - 1° Efficacy Endpoint; Events at Any Time
`
`Primary composite endpoint
`cardiovascular death
`heart failure hospitalization
`
`Subjects with events at any time**
`cardiovascular death
`heart failure hospitalization
`
`LCZ696
`
`Enalapril
`
`Hazard ratio
`
`N=4187
`n(%)
`914 (21.8)
`377 (9.0)
`537 (12.8)
`
`N=4212
`n(%)
`1117 (26.5)
`459 (10.9)
`658 (15.6)
`
`(95% CI, p-value)
`0.80 (0.73, 0.87)
`
`<0.0001
`
`558 (13.3)
`537 (12.8)
`
`693 (16.5)
`658 (15.6)
`
`0.80 (0.71, 0.89)
`0.79 (0.71, 0.89)
`
`*Analysis of time-to-first component; **Analyses of events at any time were not prospectively
`planned endpoints.
`
`
`Figure 2: PARADIGM-HF, Kaplan-Meier for 1° Efficacy Endpoint
`
`
`
`Subgroup Analyses on the 1° Endpoint:
`
`The results for the primary composite endpoint were consistent across a number of subgroups
`of interest. Some subgroup analyses were planned by the applicant; additional analyses were
`requested and/or conducted by the review team. Results were consistent (in fact, the point
`
`
`
`Reference ID: 3788923
`
`Office Director Decisional Memo – NDA 207620 – Page 11 of 29
`
`
`
`estimate of the hazard ratio was <1) for all1 of the following subgroups: age 365 YIN; age 275
`Y/N, sex, weight quartiles, race, region, US alone, NYHA functional class, eGFR 260 YIN,
`diabetes Y/N, baseline systolic blood pressure (roughly quartiles), ejection fraction (roughly
`quartiles), history of atrial fibrillation YIN, history of hypertension Y/N, prior use of ACE inhibitor
`or ARB Y/N, use of aldosterone antagonists Y/N, cause of heart failure (ischemic WM, and use
`of ICD or CRT-D. For some 434 patients enrolled in the US, where use of ICDs or CRT-D was
`common (~60% of patients), the 95% confidence interval of the hazard ratio excluded 1,
`providing some reassurance that the overall results are applicable to US. patients.
`
`Secondary Efficacy Endpoints:
`
`All—Cause Mortality:
`
`The trial was successful on its 1° endpoint; therefore, all-cause mortality was tested with the
`
`prospectively planned a (80% of the or used to test the 1° endpoint). All-cause mortality was
`statistically significantly lower in the LCZ696 arm (HR of 0.84 [0.76, 0.93]; p < 0.001)).
`It is
`important to note, however, that the finding for all-cause mortality was driven entirely by
`cardiovascular mortality (some 80% of deaths were deemed to be cardiovascular in nature;
`there were few non-cardiovascular deaths, and there were numerically more in the LCZ696
`group). Thus, the review team has concluded that statements to the effect that LCZ696
`improves survival should be accompanied by a notation that the effect was driven by a decrease
`in cardiovascular death.
`I would go further, and say that the concept of a decrease in all-cause
`mortality is actually misleading, even though the study technically “won” on this endpoint. There
`is no reasonable expectation that L02696 would decrease deaths that are unrelated to the
`cardiovascular system; indeed, there were numerically more cardiovascular deaths in the
`LCZ696 group.
`(him
`Our interest in all-cause mortality is to evaluate the unlikely possibility that a therapy has
`unanticipated, deleterious effects beyond the cardiovascular system.
`(m4)
`
`display the all-cause mortality results in Section
`14. The indication statement will state that LC2696 reduces cardiovascular mortality
`“M"
`
`KCCQ Clinical Summary Score:
`
`The change in the KCCQ Clinical Summary Score from randomization to Month 8 was tested
`
`with its prospectively planned or (20% of the or used to test the 1° endpoint). At month 8, there
`was less of a decline in the Clinical Summary Score in the LCZ696 treatment group compared
`to the enalapril group; however, the treatment effect was small. The least square mean of the
`difference was only 1.6 (95% CI [0.6, 2.7]) in subjects with a mean baseline score of ~76.
`
`Moreover, although the p-value for this analysis was quite low, it exceeded the allocated or, so
`that the study was technically not a "win” on this endpoint.
`
`Furthermore, as the reviews note, subjects who died were assigned a score of zero for all
`subsequent visits. Given the disparity in numbers of deaths between groups, deaths were
`
`1 The hazard ratio was >1 for the small number patients who were NYHA Functional Class I (about 5 % ofpatients
`in the trial). but the drug is not indicated for these patients.
`
`Reference ID: 3788923
`
`Office Director Decisional Memo — NDA 207620 — Page 12 of 29
`
`
`
`responsible for more than one-third of the treatment difference in KCCQ, providing additional
`reason to doubt the clinical significance of the effect on KCCQ.
`
`As discussed in the Clinical Review, the applicant also conducted responder analyses based on
`
`the number of subjects with 2 5—point deterioration or improvement in the clinical summary score
`f