CENTER FOR DRUG EVALUATION AND
`RESEARCH
`
`
`
`APPLICATION NUMBER:
`204569Orig1s000
`
`MEDICAL REVIEW(S)
`
`
`
`
`
`
`
`

`
`Cross Discipline Team Leader Review
`
`Cross-Discipline Team Leader Review
`
`Ronald Farkas, MD, PhD
`From
`%‘_
`
`Complete response
`
`Date of Submission
`
`PDUFA Goal Date
`Proprietary Name /
`Established
`S A
`
`names
`
`Dosae forms/Stren h
`Pro - osed Indication s
`
`August 29, 2012
`
`June 29, 2013
`Proprietary name to be determined
`Established name: suvorexant
`
`15 m,20m,30m,40m
`
`1. Introduction
`
`Orexin A and orexin B are hypothalamic neuropeptides that play a critical role in the maintenance
`of Wakefillness. Orexins are also thought to play an important role in modulation of feeding
`behavior and energy balance. Loss of orexin-containing neurons in humans is associated with
`narcolepsy, a disease characterized by excessive daytime sleepiness, cataplexy, hypnagogic and
`hypnopompic hallucinations, sleep paralysis, and other symptoms. Suvorexant antagonizes the
`two orexin receptors, orexin 1 receptor and orexin 2 receptor. Based on the involvement of orexins
`in the maintenance of wakefulness, suvorexant was developed by Merck for the treatment of
`insomnia characterized by difficulties with sleep onset and/or sleep maintenance.
`
`Suvorexant is both a new molecular entity and first-in-class.
`
`2. Background
`
`In the NDA the sponsor proposed an initial dose of 40 mg for ad11lts, and 30 mg for elderly
`patients (2 65 years of age), taken immediately before bed, with a lower dose of 20 mg for adults
`and 15 mg for elderly based on individual tolerability. However, during the review cycle, the
`sponsor informed the Division that, based on public statements made by the Division about the
`safety of zolpidem products, the sponsor was changing the dosing recommendation to initial
`dosing with the lower doses above, with the option to increase to the higher doses if necessary for
`efficacy
`
`Primary clinical review was conducted by Kachi Illoh, MD, and statistical review was conducted
`by Tristan Massie, PhD. Clinical pharmacology review was conducted by Hristina Dimova, PhD,
`and pharmacometrics review was conducted by Joo-Yeon Lee. Chemistry review was conducted
`by Akm Khairuzzaman, PhD., Biopharmaceutics review was conducted by Sandra Suarez, PhD,
`and drug substance quality aspects were reviewed by Mohan Sapru, PhD. Controlled Substance
`Staff review was conducted by Chad Reissig, PhD, and Non-Clinical review was conducted by
`Richard Siarey.
`
`Page 1 of 41
`
`Reference ID: 3331 162
`
`1
`
`

`
`Cross Discipline Team Leader Review
`
`
`
`3. CMC/Device
`Dr. Khairuzzaman found the drug product portion of the NDA to be acceptable, and without need
`for phase 4 commitments.
`
`Dr. Sapru’s review stated that with the exception of a pending issue concerning the control of
`potential genotoxic impurity
`the NDA was approvable in terms of drug
`substance.
`
`Dr. Suarez found that the NDA was acceptable from a biopharmaceutics perspective.
`
`The Office of Compliance issuance of an acceptable recommendation for drug substance
`manufacturing and testing facilities was pending at the time of this review.
`4. Nonclinical Pharmacology/Toxicology
`Dr. Richard Siarey completed the primary nonclinical review, and Dr. Lois Freed completed a
`supervisory memo.
`
`Dr. Siarey’s overall conclusion was that from a nonclinical perspective, approval of the suvorexant
`NDA was recommended. However, he found evidence that catapelxy was observed in dogs
`exposed to MK-4305 (suvorexant) near Tmax, although he concluded that additional information
`could have been gained by studying the drug in an experimental model that has been used for
`diagnosing cataplexy in dogs. Dr. Siarey suggested that since cataplexy occurred in dogs near
`Tmax, a time at which if used for insomnia patients would ordinarily be in bed, safety concern for
`humans was reduced. Dr. Siarey also found that the neurobehavioral assessment in the pre- and
`post-natal developmental study was not complete, as the passive avoidance tests was performed
`too early in development, while learning/acquisition tests and retention/memory tests were not
`conducted. He recommended that these studies be repeated/conducted.
`
`Dr. Freed agreed that nonclinical studies suggested a theoretical safety concern for cataplexy,
`concluding that clinical implications, if any, are an issue for the clinical team to decide. She
`suggested that findings of cataplexy in dog be described in labeling, but would not require
`additional nonclinical studies of cataplexy, and concluded that the neurobehavioral assessments
`were sensitive enough to detect at least some adverse effects, and since none were observed, the
`studies were minimally acceptable. Therefore, the nonclinical team had no suggested post-
`marketing requirements.
`
`
`5. Clinical Pharmacology/Biopharmaceutics
`
` single Clinical Pharmacology review combined the findings of Dr. Dimova and Dr. Lee.
`
` A
`
`
`
`Half life and accumulation
`
`Page 2 of 41
`
`Reference ID: 3331162
`
`2
`
`(b) (4)
`
`

`
`Cross Discipline Team Leader Review
`
`The half-life of suvorexant is about 12 hours, such that levels accumulate to steady state over
`several days of dosing.
`
`Figure 1: Suvorexant PK, 40 mg
`
`
`
`
`CDTL: Figure 1 illustrates how the long half-life of suvorexant may impact both safety and
`efficacy. Suvorexant blood level about 12 hours after a single dose is about the same as the
`Cmax of a single 10 mg dose (blue dashed line in figure). As discussed below in Section 7, the
`10 mg dose appears to be effective for inducing and maintaining sleep. While circadian
`effects might make patients less sensitive to somnolence during the day versus at night, it is
`concerning that ‘effective’ levels are present during the day. With repeat dosing, daytime
`levels increase due to drug accumulation, such that suvorexant levels about 12 hours after
`the previous night’s dose are similar to Cmax from the 15 mg dose, again a dose found to be
`effective for sleep latency and maintenance.
`
`The long half-life also is likely to affect efficacy, particularly for sleep latency, which is more
`dependent than sleep maintenance on the time between dosing and blood levels reaching an
`effective level. At the first dose, suvorexant blood level must go from zero to some level
`before the drug could be effective. However, with chronic dosing of 40 mg, suvorexant blood
`level at bedtime, before taking that night’s dose, is already about the same as Cmax from the
`10 mg dose, a dose seemingly effective for sleep latency. This same relationship holds for any
`dose (with the steady-state suvorexant level proportional to the dose), such that potentially
`even if a low dose (10 mg or even lower) is less effective on night 1 for sleep latency than a
`high dose, on subsequent nights of chronic dosing, accumulation of suvorexant would allow
`suvorexant blood levels to more quickly reach an effective level. The difference in efficacy
`between low and high dose would diminish because, while the high dose would also lead to
`accumulation of suvorexant, the exposure from the high dose would already have been in or
`near the plateau region of the dose-response relationship, such that higher exposure would
`lead to little if any greater efficacy.
`
`Page 3 of 41
`
`Reference ID: 3331162
`
`3
`
`

`
`Cross Discipline Team Leader Review
`
`
`reference
`
`
`
`Single Dose
`
`non-obese male
`
`
`
`
`test
`
`
`Obese*
`male
`
`
`Intrinsic factors
`Table 1 shows effect of gender and BMI on suvorexant exposure. Exposure is increased in obesity
`and in women compared to men.
`
`Table 1: Gender and BMI
`
`
`
`
`
`
`
`
`ratio of
`
`exposure metric
`
`
`
`1.1
`4
`1.0
`7
`1.0
`3
`
`1.1
`8
`1.0
`9
`1.0
`4
`
`1.3
`9
`1.1
`2
`1.1
`
`1.4
`5
`1.1
`7
`1.1
`2
`
`
`
`
`
`
`
`AUC
`
`Cmax
`
`C9 hours
`
`
`AUC
`
`Cmax
`
`C9 hours
`
`
`
`Cmax
`C9 hours
`
`
`AUC
`
`Cmax
`
`C9 hours
`
`
`
`
`
`1.17
`1.09
`1.04
`
`
`
`
`
`non-obese
`female
`
`
`
`
`
`
`
`
`
`
`
`obese
`female
`
`
`
`
`
`
`
`non-obese male obese male AUC
`
`
`
`
`
`
`
`
`
`
`
`
`
`Multiple
`Dose
`
`
`
`
`
`
`
`
`
`
`
`
`non-obese
`female
`
`
`
`
`
`
`
`Overall female vs male
`
`
`
`
`
`
`
`*Definition of obese (>30 kg/m2)
`
`
`
`
`
`obese
`female
`
`
`
`
`
`
`AUC
`Cmax
`C9
`
`Page 4 of 41
`
`Reference ID: 3331162
`
`4
`
`

`
`Cross Discipline Team Leader Review
`
`CDTL: Increased exposure in women vs. men and obese vs. non-obese patients are
`potentially clinically important. The clinical pharmacology review recommended a 50%
`lower (e.g. 10 mg) starting dose in obese females based on the approximate doubling of
`suvorexant exposure vs. non-obese males, and the approximate 20% increase in blood levels
`the morning after dosing at steady state.
`
`If the initial recommended dose is 10 mg with the option of increasing the dose if clinically
`indicated, it is not clear that specific dose adjustment needs to be made for these differences;
`instead, labeling indicating that such exposure differences occur, and should enter into
`clinical judgments about dose escalation, appears adequate.
`
`Age
`Elderly patients were predicted by the sponsor to have about 15% higher blood levels 9 hours after
`dosing versus non-elderly patients based on a combination of covariate effects of age, BMI, and
`creatinine clearance. However, Dr. Dimova found that data from phase 1 studies suggested that age
`alone did not have an effect on suvorexant PK.
`CDTL: The Clinical Pharmacology review concludes dose-adjustment for age is not
`necessary. I agree.
`
`CYP3A inhibition
`Exposure (AUC) of suvorexant is increased about 3-fold by strong CYP3A inhibitors and about 2-
`fold by moderate CYP3A inhibitors (Figure 2).
`
`
`
`Figure 2: Effect of CYP3A Inhibition on Suvorexant Exposure
`
`The Clinical Pharmacology review concluded the following regarding CYP3A:
`
`Page 5 of 41
`
`Reference ID: 3331162
`
`
`
`5
`
`

`
`Cross Discipline Team Leader Review
`
`o Exposure is increased about 3-fold by strong CYP3A inhibitors, and suvorexant should
`not be co-administered with strong CYP3A inhibitors
`o Exposure is increased about 2-fold by moderate CYP3A inhibitors, and dosing should
`be adjusted accordingly
`o Efficacy may be reduced by CYP3A inducers
`
`
`CDTL: I agree with the recommendations regarding CYP3A inhibitors; with 10 mg dosing
`and strong CYP3A inhibotors, exposure would be similar to that from high dose suvorexant
`(30 mg/ 40 mg), which this review concludes is not safe, while exposure from the 10 mg dose
`used with moderate inhibitors would be similar to that from 20 mg suvorexant, which this
`review concludes is adequately safe so long as patients refrain from driving.
`
`Exposure vs. Somnolence
`Analysis from combined controlled trials showed that adverse events of somnolence were
`exposure-related (Figure 3).
`
`Figure 3: Suvorexant Exposure vs. Incidence of Somnolence
`
`
`
`
`CDTL: Next-day suvorexant blood levels in adults after nighttime administration of the 15
`mg dose overlap with blood levels from the 20 mg dose that caused driving impairment in the
`formal driving study (not shown), as discussed in Section 8. The observation above that
`exposure and somnolence are correlated supports concern that patients at the higher end of
`the population distribution of exposure from the 15 mg dose are more likely to be at similar
`risk of adverse effects to patients with similar exposure from the 20 mg dose. As discussed in
`Section 7, a starting dose of 10 mg would likely be effective for many patients who would be
`at the high end of exposure, thus reducing risk of driving impairment while not diminishing
`efficacy for these patients.
`
`
`
`Page 6 of 41
`
`Reference ID: 3331162
`
`6
`
`

`
`Cross Discipline Team Leader Review
`
`6. Clinical Microbiology
`Not applicable
`7. Clinical/Statistical- Efficacy
`The sponsor conducted two similarly designed phase 3 efficacy trials of 3 months duration, study
`028 and 029. The studies enrolled both adult patients 18- to 64 years of age, and elderly patients
`age 65 and above. About 40% of the subjects in each study were elderly. Two doses were tested in
`each age group: for adults, 40 mg or 20 mg suvorexant, and for elderly patients, 30 mg or 15 mg.
`Randomization was to low dose, high dose, or placebo in a 2:3:3 ratio, respectively. The 40 mg
`and 30 mg ‘high’ doses (HD) in adults were expected to result in similar exposure, as were the 20
`mg and 15 mg ‘low’ doses (LD), and for analysis the high doses were pooled and the low doses
`were pooled. The studies were powered for the high dose sleep maintenance endpoints.
`
`CDTL: As noted in Dr. Massie’s review, the sponsor’s goals for the phase 3 studies were
`ambitious, in that they aimed to demonstrate effects for both elderly and non-elderly patients
`on both sleep maintenance and sleep onset, in terms of both an objective and a subjective
`assessment for each, in the same study. The sponsor also powered the studies for the high
`dose (the low dose had 30- to 40% less patients), and the multiplicity method tested the high
`dose first. The studies were thus underpowered for the low dose, such that non-statistically
`significant findings for some endpoints at some time points for the low dose is unsurprising,
`and in no way provides interpretable evidence against efficacy. Likewise, the phase 2 study
`was powered for objective sleep maintenance, but not latency or subjective sleep endpoints,
`such that non-statistically significant findings can not be taken as meaningful evidence of
`lack of efficacy at lower suvorexant doses. The question, discussed below, then becomes if
`enough evidence has been provided in the overall database for appropriate dose-selection,
`even though studies were underpowered to provide this data in the form of consistently
`statistically positive endpoints.
`
`Patients used suvorexant immediately before bedtime when at home, but during inpatient
`polysomnography nights, suvorexant was dosed 30 minutes before lights out and initiation of
`recording.
`
`Sleep maintenance, phase 3 studies
`Objective evidence of benefit for sleep maintenance, as measured by Polysomnographic (PSG)
`Wake After Sleep Onset (WASO) was positive in both study 28 and 29 at each time point (night 1,
`month 1, and month 3) for both high dose (HD)(30 and 40 mg combined) and low dose (LD)(15
`and 20 mg combined)(Table 2, Table 3).
`
`
`
`Table 2: Study 28 PSG-WASO
`Night 1
`
`
`HD vs Pbo
`
`
`LD vs Pbo
`
`
`
`Month 1
`HD vs Pbo
`
`
`
`
`
`
`
`
`
`
`
`Difference (min)
`-38.4
`
`-32.5
`
`
`-26.3
`
`
`
`
`
`
`
`
`
`
`
`p-value
`<0.00001
` <0.00001
`
`<0.00001
`
`Page 7 of 41
`
`Reference ID: 3331162
`
`7
`
`

`
`Cross Discipline Team Leader Review
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`-26.4
`
`-22.9
`-16.6
`
`
`
`
`
`
`Difference (min)
`-42.0
`
`-37.0
`
`
`-29.4
`-24.1
`
`-29.4
`-31.1
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
` <0.00001
`
`<0.00001
` 0.000009
`
`
`
`p-value
`<0.00001
` <0.00001
`
`<0.00001
` <0.00001
`
`<0.00001
` 0.000009
`
`LD vs Pbo
`
`Month 3
`HD vs Pbo
`LD vs Pbo
`
`
`
`
`
`
`
`
`
`
`
`
`
`Table 3: Study 29 PSG-WASO
`Night 1
`
`
`HD vs Pbo
`
`
`LD vs Pbo
`
`
`
`Month 1
`HD vs Pbo
`LD vs Pbo
`
`Month 3
`HD vs Pbo
`LD vs Pbo
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Subjective perception of minutes awake at night, as measured by subjective WASO, was positive
`for most time points, but not month 3 for LD in study 28, or week 1 for LD in study 29 (Table 4,
`Table 5).
`
`Table 4: Study 28, Subjective WASO
`Week 1
`
`
`
`HD vs Pbo
`
`
`
`LD vs Pbo
`
`
`
`
`Month 1
`HD vs Pbo
`LD vs Pbo
`
`Month 3
`HD vs Pbo
`LD vs Pbo
`
`Difference (min)
`-10.5
`
`-6.8
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`-9.5
`-5.4
`
`-6.9
`-2.4
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`p-value
`<0.00001
` 0.003
`
`
`
`
`
`
`0.00025
`0.06
`
`0.006
`0.39
`
`
`
`Table 5: Study 29, Subjective WASO
`Week 1
`
`
`
`HD vs Pbo
`
`
`
`LD vs Pbo
`
`
`
`
`Month 1
`HD vs Pbo
`LD vs Pbo
`
`
`
`
`
`
`
`
`
`
`Difference (min)
`-8.4
`
`
`-4.2
`
`
`
`-8.7
`-8.4
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`p-value
`0.0005
` 0.13
`
`0.001
`0.006
`
`
`
`
`
`
`Page 8 of 41
`
`Reference ID: 3331162
`
`8
`
`

`
`Cross Discipline Team Leader Review
`
`
`
`
`
`
`
`
`
`
`-8.9
`-7.7
`
`
`
`
`
`
`
`
`
`
`0.002
`0.02
`
`
`
`
`
`
`
`
`Month 3
`HD vs Pbo
`LD vs Pbo
`
`
`
`
`
`
`
`Sleep latency, phase 3 studies
`Objective evidence of benefit for sleep latency, as measured by PSG-LPS, was positive in study 28
`for each time point, for both HD and LD, but in study 29, while PSG-WASO was positive for each
`time point for HD, it was positive for LD only at night 1 (Table 6, Table 7).
`
`Table 6: Study 28, LPS
`Night 1
`
`HD vs Pbo
`
`LD vs Pbo
`
`
`Month 1
`HD vs Pbo
`LD vs Pbo
`
`Month 3
`HD vs Pbo
`LD vs Pbo
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Difference (min)
`-10.3
`
`-9.6
`
`
`
`-11.2
`-10.3
`
`-9.4
`-8.1
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`p-value
`0.00002
`0.0004
`
`0.00002
`0.0004
`
`0.0004
`0.0061
`
`
`
`
`
`
`
`
`
`Table 7: Study 29, LPS
`Night 1
`
`HD vs Pbo
`
`LD vs Pbo
`
`
`Month 1
`HD vs Pbo
`LD vs Pbo
`
`Month 3
`HD vs Pbo
`LD vs Pbo
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Difference (min)
`-21.7
`
`-12.4
`
`
`
`-12.1
`-7.8
`
`
`-3.6
`-0.3
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`p-value
`<0.00001
` 0.004
`
`0.00004
`0.03
`
`0.27
`0.9
`
`
`
`
`
`
`
`
`
`
`Subjective perception of sleep latency, as measured by subjective time sleep onset (sTSO), was
`positive in study 28 and 29 for each time point for HD, but negative for each time point for LD
`(although nominal p-values were about 0.05 or less for all LD time points)(Table 8, Table 9).
`
`Table 8: Study 28, Subjective Sleep Latency (sTSO)
`Week 1
`
`
`
`Difference (min)
`HD vs Pbo
`
`
`
`-5.7
`
`
`LD vs Pbo
`
`
`
`-5.6
`
`
`
`
`
`
`
`
`
`p-value
`0.0061
` 0.016
`
`Page 9 of 41
`
`Reference ID: 3331162
`
`9
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`0.003
`0.052
`
`0.0002
`0.04
`
`
`
`p-value
`<0.00001
` 0.006
`
`0.00003
`0.05
`
`0.00003
`0.04
`
`Cross Discipline Team Leader Review
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`-7.4
`-5.4
`
`-8.4
`-5.2
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Month 1
`HD vs Pbo
`LD vs Pbo
`
`Month 3
`HD vs Pbo
`LD vs Pbo
`
`
`
`Table 9: Study 29, Subjective Sleep Latency (sTSO)
`Week 1
`
`
`
`Difference (min)
`HD vs Pbo
`
`
`
`-13.1*
`
`LD vs Pbo
`
`
`
`-7.5
`
`
`
`Month 1
`HD vs Pbo
`LD vs Pbo
`
`Month 3
`HD vs Pbo
`LD vs Pbo
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`-12.8*
`-6.9
`
`
`-13.2*
`-7.6
`
`
`
`
`
`
`
`
`
`
`Dr. Massie’s review analyzed results for elderly and adult separately, pooling results for study 28
`and 29 (Table 10)(the sponsor was not required to demonstrate statistical significance separately in
`the two age groups, so only point estimates are shown in the table). The effects in adult and elderly
`patients are generally similar.
`
`
`
`Table 10: Efficacy by Age (Adult, Elderly)
`
`Adult Elderly
`
`Adult
`
`Elderly
`
`Low Dose
`(min)
`
`
`
`PSG-WASO
` Day 1
` Day 30
` Day 90
`
`Subjective total sleep time
` Week 1
` Month 1
` Month 3
`
`PSG-LPS
`
`
`-27
`-26
`-18
`
`
`13
`20
`13
`
`
`
`
`-40
`-27
`-15
`
`
`17
`16
`19
`
`
`
`High Dose
`(min)
`
`
`-33
`-26
`-27
`
`
`23
`24
`23
`
`
`
`
`-45
`-26
`-17
`
`
`25
`21
`21
`
`
`
`Page 10 of 41
`
`Reference ID: 3331162
`
`10
`
`

`
`Cross Discipline Team Leader Review
`
` Day 1
` Day 30
` Day 90
`
`Subjective latency (sLSO)
` Week 1
` Month 1
` Month 3
`
`-12
`-12
`-3
`
`
`-6
`-7
`-6
`
`-10
`-5
`-6
`
`
`-7
`-4
`-7
`
`-15
`-15
`-5
`
`
`-9
`-11
`-12
`
`-18
`-7
`-8
`
`
`-9
`-9
`-9
`
`
`
`Phase 2 Study 006
`The sponsor also conducted a phase 2 dose-finding two-period cross-over study that examined the
`efficacy of 10 mg, 20 mg, 40 mg, and 80 mg suvorexant. A total of 254 patients were randomized,
`with about 60 patients in each dose arm. PSG-WASO was statistically significantly reduced
`(p<0.001) for all doses, with the following point estimates (Table 11):
`
`
`Table 11: PSG-WASO, Study 006
`Dose
`
`Night 1 (min)
`10 mg
`
`-21
`
`20 mg
`
`-25
`
`40 mg
`
`-34
`
`80 mg
`
`-37
`
`
`
`Week 4 (min)
`
`-21
`
`-28
`
`-33
`
`-29
`
`
`
`
`In contrast, by the multiplicity strategy no dose was positive for PSG-LPS, with following point
`estimates and nominal p-values (Table 12):
`
`Table 12: PSG-LPS, Study 006
`Dose
`
`Night 1 (min) p-value
`10 mg
`
`-3
`0.6
`
`20 mg
`
`-9
`.13
`
`40 mg
`
`-23
`<0.001
`80 mg
`
`-25
`<0.001
`
`Week 4 (min) p-value
`
`-2
`0.6
`
`-22
`<0.001
`
`-4
`0.5
`
`-10
`0.07
`
`
`
`Dr. Massie’s review found evidence of a carryover effect between the first- and second study
`periods, and therefore performed an analysis of LPS restricted to period 1. Applying the pre-
`specified multiplicity testing strategy to this alternative analysis, the 80 mg dose significantly
`improved LPS, and the nominal p-values were <0.05 for all other comparisons of drug vs. placebo
`except for 40 mg at week 4 (p-value 0.06), with point estimates as follows:
`
`
`
`Dose
`10 mg
`20 mg
`40 mg
`80 mg
`
`Night 1 (min) p-value
`
`-19
`0.02
`
`
`-17
`.03
`
`
`-31
`<0.001
`
`-22
`0.007
`
`
`Week 4 (min) p-value
`
`-20
`0.02
`
`-25
`0.003
`
`-16
`0.06
`
`-20
`0.02
`
`Page 11 of 41
`
`Reference ID: 3331162
`
`11
`
`

`
`Cross Discipline Team Leader Review
`
`
`
`Dr. Massie’s review also found unexpected variability between placebo groups for the different
`doses tested, and that results for the 10 mg dose were particularly sensitive to the placebo group
`involved in the comparison. Based on pooled placebo, the 10 mg dose was nominally positive for
`PSG-LPS at both day 1 and week 4 (p = 0.01 for each).
`
`Dr. Massie’s review found little statistical evidence for efficacy of the 10 mg dose on subjective
`endpoints in study 006 beyond results that were directionally consistent with efficacy. The overall
`conclusion was that the application overall provided clear evidence for the efficacy of suvorexant
`for sleep maintenance, and weaker evidence for efficacy for sleep latency, citing the failure to
`reach statistical significance at the 3-month time point for PSG-LPS. He also concluded that study
`006 provided a suggestion of efficacy for 10 mg, particularly for WASO, but notes that no other
`study examined 10 mg to provide replication of findings, and that efficacy of 10 mg was not
`studied in elderly patients. He concluded that if the phase 3 doses (15 through 40 mg) are
`considered to have too much risk of next-day driving impairment then another study of the 10 mg
`dose may be needed.
`
`Dr. Illoh’s review found substantial evidence of efficacy for the high doses only (30 and 40 mg). It
`notes that suvorexant high dose failed to significantly improve LPS at month 3 in trial 29, but finds
`subjective endpoints to be more important than objective because they measure patient perception
`not a biomarker, such that positive findings for subjective time to sleep onset provide in study 29
`provide, in combination with positive findings in study 28, adequate evidence of efficacy. For the
`low doses, the review notes that because in study 28, low doses did not improve sleep onset
`endpoints beyond month 1, efficacy of low doses for sleep onset remains uncertain. In another
`section, the review states that suvorexant low dose may be considered in sleep maintenance
`insomnia, but its benefit for sleep onset insomnia remains uncertain.
`
`CDTL Discussion:
`
`Efficacy
`To support a claim for insomnia the Division required that efficacy be demonstrated in at
`least two studies of 3-months duration, in both adult and elderly patients. The 3-month
`duration was intended to be a sufficient duration to demonstrate efficacy for chronic
`insomnia, which is often treated with drugs for 3 months or longer.
`
`Due to concern about possible age-related differences in efficacy and safety of drugs for
`insomnia, the sponsor was expected to provide enough evidence to allow the Division to
`conclude that suvorexant was safe and effective in both adult and elderly patients. The
`Division did not, however, require separate positive studies for adult and elderly patients.
`
`For a claim in sleep latency, the Division required positive findings on both an objective
`polysomnographic measure of sleep latency, latency to persistent sleep (LPS), and the
`patient’s subjective estimation of the time taken to fall asleep, sleep onset latency (SOL). For
`a claim in sleep maintenance, the Division required positive findings on both an objective
`polysomnographic measure of sleep maintenance, wake after sleep onset (WASO), and the
`
`Page 12 of 41
`
`Reference ID: 3331162
`
`12
`
`

`
`Cross Discipline Team Leader Review
`
`patient’s subjective estimation of wake after sleep onset, subjective WASO (sWASO). The
`objective and subjective endpoints were required to demonstrate, respectively, that the drug
`did, in fact, treat insomnia through positive effect on sleep latency and/or maintenance, and
`that this positive effect was large enough to be apparent to the patient, as a marker of clinical
`meaningfulness.
`
`While subjective estimation of sleep time has been an endpoint required by the Division, it is
`known to be an inaccurate reflection of objective sleep time, and thought to be even less
`reflective of objective benefit due to psychoactive effects of insomnia drugs themselves,
`including amnestic effects. Thus, there is even concern that sleep might be misperceived due
`to what would ordinarily be considered an adverse drug effect (amnesia). Thus, there is
`increased realization that while subjective estimates of sleep remain useful markers, they
`should be interpreted with these cautions in mind. Historically, no requirement has been set
`by the Division for either objective or subjective benefit for daytime function, or more global
`patient reported outcomes for insomnia, like the Insomnia Severity Index. However,
`particularly in the context of increased FDA focus on patient perspective on disease
`symptoms, the Division is putting increasing emphasis on such endpoints because they
`appear to be clearly clinically meaningful to patients.
`
`The discussion below of efficacy evidence is based in particular on situations when
`substantial evidence of efficacy can be established based on less than two positive studies
`through reliance on conclusive findings of efficacy in related settings, as described in the
`FDA Guidance for evidence of effectiveness1.
`
`For sleep maintenance, the high dose arms of suvorexant were effective for both objective
`and subjective endpoints, in two studies, with efficacy maintained through the full 3 month
`study duration. With these findings suvorexant meets the basic legal standard for quantity
`(and quality) of evidence to support that the drug is effective for sleep maintenance.
`
`For objective sleep latency, HD was positive through the full 3 month study duration in study
`28, while study 29 was positive for night 1 and day 30. Subjective sleep latency was positive
`for HD through the full 3 month duration in both study 28 and 29. Sleep latency and sleep
`maintenance are closely related indications; sleep maintenance in many respects can be
`considered as difficulty with sleep latency when trying to fall back asleep after middle-of-the-
`night awakenings. Given the demonstrated efficacy of suvorexant in sleep maintenance, it is
`therefore reasonable to accept a single additional study positive for both objective and
`subjective endpoints, study 29, as adequate support for the sleep latency claim.
`
`The point estimates are similar for the effect of suvorexant on adult versus elderly patients
`(Table 10), and from this it is reasonable to conclude that study 28 and 29 provide enough
`evidence to conclude that suvorexant is effective in both adult and elderly patients, for both
`sleep onset and sleep maintenance.
`
`Dose/response
`
`
`1 Guidance for Industry: Providing Clinical Evidence of Effectiveness for Human Drug and Biological Products
`
`Page 13 of 41
`
`Reference ID: 3331162
`
`13
`
`

`
`Cross Discipline Team Leader Review
`
`While insomnia is associated with an increased incidence of long-term adverse health
`outcomes, data is currently lacking that would justify using a higher (and less safe) drug dose
`for insomnia than necessary to satisfactorily treat current symptoms. The Guidance for
`evidence of effectiveness cited above notes that information about effectiveness of one dose is
`relevant to the effectiveness of other doses, and describes a flexible approach to determining
`efficacy for different doses based on not only additional clinical efficacy data but also
`pharmacokinetic and dose/response data, sometimes even in the absence of clinical efficacy
`data.
`
`In the phase 3 trials (28 and 29), the combined low dose arms were positive at all time points
`for PSG-WASO in both studies, and for subjective WASO at all time points in study 29.
`Given the demonstrated effectiveness of suvorexant for sleep maintenance (at the high doses),
`this additional data (more than one additional positive study) is thus more than sufficient to
`conclude efficacy of the lower doses for sleep maintenance.
`
`For the low doses, PSG-LPS was positive at all time points in study 28, but only night 1 in
`study 29. Subjective sleep latency was positive at week 1 in both studies, but at month 1 and
`month 3, in both studies, the p-values were only nominally positive (0.04 in both studies at
`month 3). In the context of positive findings for the high dose, positive findings in only one
`study would be necessary to support efficacy of the low dose. Objective findings are provided
`by study 28. For the subjective endpoint, it seems reasonable to consider two nominally
`positive studies to be of similar persuasiveness as a single study positive on a pre-specified
`positive endpoint, as replication provides important, qualitatively different, reassurance in
`addition to the p-value. Thus, I conclude that the phase 3 studies adequately support the
`efficacy of doses as low as 15 mg for chronic sleep latency and/or sleep maintenance.
`
`Efficacy data is available for the 10 mg dose from study 006, the phase 2 dose finding study.
`Study 006 was positive for the 10 mg dose by the pre-specified analysis for objective sleep
`maintenance, and as described by Dr. Massie’s review by reasonable post-hoc sensitivity
`analyses for objective sleep latency. Efficacy of the 10 mg dose is also supported by the
`exposure/response analysis conducted by Dr. Lee on data from the pivotal studies 28 and 29.
`She found no clear evidence of decreased efficacy for objective sleep maintenance or latency
`in patients who had suvorexant exposures (after dosing with 15 mg or higher dosage forms)
`similar to exposures that, from PK studies, occur after the 10 mg dose.
`
`Dr. Massie’s review found that for the 10 mg and 20 mg doses there was little evidence from
`study 006 of efficacy on subjective total sleep time or subjective time to sleep onset, although
`subjective time to sleep onset was nominally positive (p = 0.03) at week 4. Some evidence for
`subjective benefit from the 10 mg (and 20 mg) dose is provided by the Insomnia Severity
`Index (ISI), a patient-reported outcome of objective complaints in insomnia consisting of
`seven questions including patient satisfaction and worry about sleep, and perception of how
`insomnia interferes with their functioning. Figure 4 shows ISI results for study 006. Benefit
`on ISI appeared to be similar for all doses, noting however that the p-value was not
`nominally positive for 10 mg.
`
`
`
`Page 14 of 41
`
`Reference ID: 3331162
`
`14
`
`

`
`Cross Discipline Team Leader Review
`
`
`
`Figure 4: Insomnia Severity Index, Study 006, Period 1
`
`
`Figure 4 legend: The correlation between ISI total score and insomnia severity is generally considered to be the
`following: 0-7, no clinically significant insomnia; 8-14, subthreshold insomnia; 15-21, moderate insomnia; 22-28,
`severe insomnia.
`
`As also discussed in section 5, based on pharmacokinetics, many patients taking the 10 mg
`dose would have exposure to suvorexant as higher or higher that from the 15 mg or even 20
`mg dose, including obese females, a key demographic that would use suvorexant if approved,
`and who have exposure (AUC) about 1.5-fold higher than non-obese men. This supports
`efficacy of the 10 mg dose based on PK.
`
`More generally, as shown in Figure 6, because of variability across the population in
`exposure to a given dose, there is considerable overlap of suvorexant exposure among
`
`Page 15 of 41
`
`Reference ID: 3331162
`
`15
`
`

`
`Cross Discipline Team Leader Review
`
`subjects receiving the 10 and 20 mg doses. Thus, logically, if the 20 mg dose is effective, then
`many patients at the higher end of the distribution of exposures after receiving the 10 mg
`dose (perhaps the top half of patients) will have to a similarly effective exposure to
`suvorexant. The abov