From: [email protected] To: [email protected] Subject: Condition Petition for Thomas Rosenberger Date: Tuesday, December 31, 2019 5:41:25 PM Attachments: Question 1 (3) (1).pdf Question 2 Insomnia (3) (1).pdf Question 3 Insomnia (6) (2).pdf Question 5 Insomnia (3).pdf

This message was sent from the Condition page on medicalmarijuana.ohio.gov.

Box was check regarding file size being too large to upload. Action needed!

Name: Thomas Rosenberger Address: 815 Grandview Ave Suite 400, Columbus, OH, 43215 Phone: (614) 706-3782 Email: [email protected]

Specific Disease or Condition: Insomnia

Information from experts who specialize in the disease or condition. See attached file

Question 1 Insomnia (3) (1).pdf

Relevant medical or scientific evidence pertaining to the disease or condition. See attached file

Question 2 Insomnia (3) (1).pdf

Consideration of whether conventional medical therapies are insufficient to treat or alleviate the disease or condition. See attached file

Question 3 Insomnia (6) (2).pdf

Evidence supporting the use of medical marijuana to treat or alleviate the disease or condition, including journal articles, peer-reviewed studies, and other types of medical or scientific documentation. File larger than 3MB

Letters of support provided by physicians with knowledge of the disease or condition. This may include a letter provided by the physician treating the petitioner, if applicable. See attached file

Question 5 Insomnia (3).pdf

Question 1

Information from experts who specialize in the disease or condition

Contents

Overview – 3

Clinical Pharmacology in Sleep Medicine – 4

Overview

Insomnia is a difficult condition to treat in part because it both a condition in and of itself and a symptom of other conditions. This makes it difficult for physicians to determine the correct treatment options.

Choosing the correct treatment option is complicated by the myriad of severe side effects pharmaceuticals intended to help a patient sleep can cause. Seizure, depression and memory issues are just a few.

“Clinical Pharmacology in Sleep Medicine” attempts to clarify the risks of various medications intended to help with insomnia and other sleep disorders. International Scholarly Research Network ISRN Pharmacology Volume 2012, Article ID 914168, 14 pages doi:10.5402/2012/914168

Review Article Clinical Pharmacology in Sleep Medicine

Ashley Proctor and Matt T. Bianchi

Sleep Division, Neurology Department, Massachusetts General Hospital, Wang 720, Boston, MA 02114, USA

Correspondence should be addressed to Matt T. Bianchi, [email protected]

Received 30 April 2012; Accepted 7 June 2012

Academic Editors: T. Kumai, M. van den Buuse, and R. Villalobos-Molina

Copyright © 2012 A. Proctor and M. T. Bianchi. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The basic treatment goals of pharmacological therapies in sleep medicine are to improve waking function by either improving sleep or by increasing energy during wakefulness. Stimulants to improve waking function include amphetamine derivatives, modafinil, and caffeine. Sleep aids encompass several classes, from to over-the-counter . Other medications used in sleep medicine include those initially used in other disorders, such as epilepsy, Parkinson’s disease, and psychiatric disorders. As these medications are prescribed or encountered by providers in diverse fields of medicine, it is important to recognize the distribution of adverse effects, drug interaction profiles, metabolism, and cytochrome substrate activity. In this paper, we review the pharmacological armamentarium in the field of sleep medicine to provide a framework for risk-benefit considerations in clinical practice.

1. Introduction and idiopathic hypersomnia are also treated primarily with stimulants. Given the high prevalence of sleep complaints in the general Insomnia can be considered a constellation of symptoms population and in patients with a variety of comorbid dis- with a variety of underlying causes [6]. As a symptom, it orders, the pharmacological treatment options for sleep dis- can be secondary to disorders of mood, pain, or a variety of orders are common considerations for sleep specialists and other neurological and general medical disorders. It can be nonspecialists alike [1–4]. Clinical pharmacology in sleep primary in the sense that it exists in the absence of other medicine can be loosely classified into drugs aimed at treat- identifiable causes, such as insomnia from psychophysio- ing sleepiness, sleeplessness, and sleep-related movements. logical associations, or secondary to a number of other Although most of these are available by prescription only, the medical and psychiatric issues [7–13]. One of the most stimulant caffeine and the intriguing yet poorly understood aspects of insomnia is the are common over-the-counter options for sleepiness and misperception phenotype, in which patients underestimate sleeplessness, respectively. their sleep times compared to objective measurements [14]. The primary hypersomnias are uncommon compared to Insomnia can also be the presenting feature of circadian disorders that include sleepiness as a secondary symptom to phase disorders—most commonly delayed circadian phase sleep disruption [5]. When presented with the patient report- [15]. The primary challenge in regards to the diagnosis and ing sleepiness, it is critical to investigate potential primary treatment of insomnia is that both depend entirely on the causes, such as sleep apnea or insomnia. Pain syndromes, clinical history, with no basis in objective testing. mood disorders, and general medical problems may be Restless leg syndrome and periodic limb movements of comorbid with sleep apnea and/or disrupted sleep. However, sleep are the most common movement disorders resulting in residual daytime symptoms persist in some patients despite sleep disturbance [16, 17]. The former is a strictly clinical optimized management of potential primary causes, leading diagnosis, while the latter is a polysomnographic finding. to consideration of stimulant agents in the appropriate Both are treated similarly, often beginning with interrogation clinical setting. Primary hypersomnias such as narcolepsy of iron stores and oral repletion when needed, followed by 2 ISRN Pharmacology dopaminergic medications, as well as off-label use of other which include UpToDate and Micromedex. Interested read- classes. REM behavior disorder is most commonly treated ers seeking further detail are directed to these sources. with [18]. The purpose of this paper is to provide an overview of the 2. Insomnia medications most commonly encountered in sleep medicine. It is not intended as a clinical guideline and prescription Multiple drug treatments exist for the treatment of insomnia, decisions should be undertaken with appropriate expertise, ranging from dedicated hypnotics, such as , to consultation, and consideration of available information multifunctional benzodiazepines [3, 22]. Other drugs have about adverse effects, interactions, and safety issues. The been used for insomnia on account of drowsiness as a organization of tables and figures includes basic information prominent side effect such as antihistamines and , abouteachdrugsuchashalf-life,excretion(renalorhepatic), as well as over-the-counter use of the hormone pregnancy class and lactation considerations, and inter- [23]. In the United States, the Food and Drug Administration actions with food, herbals, and smoking (Tables 1–4). In has approved the ligands (zolpidem, Table 1, the generic names of a variety of sleeping pills are , and ), , , and certain shown along with the biological half-life, primary mode benzodiazepines. Other sleep aids shown in this paper of excretion, pregnancy class, presence in breast milk, and are considered off-label. The use of drug treatments is interactions with food, herbs, and smoking tobacco. The recommended only in the short term, meaning usually 7– FDA defines pregnancy categories as follows: (A) adequate 10 nights. Long-term pharmacological treatment is not and well-controlled studies failed to demonstrate fetal risk suggested, and only the newer agents such as zolpidem have in the first trimester (and there is no evidence of risk in even been studied in clinical trials lasting more than a few later trimesters); (B) animal reproduction studies have failed weeks. Despite this practice recommendation, patients with to demonstrate fetal risk and there are no adequate and chronic insomnia may not achieve spontaneous, behavioral well-controlled studies in pregnant women, or human data modification, or drug-assisted remission, and appropriate reassuring despite animal studies showing risk; (C) animal treatment in these cases remains uncertain. reproduction studies are either not available or have shown The main nonpharmacological approaches span the gen- adverse fetal effects and there are no adequate and well- eral categories of (1) treating the underlying contributors controlled studies in humans, but potential benefits may (such as pain or mood disorders), (2) optimizing sleep hy- warrant use of the drug in pregnant women despite potential giene, and (3) cognitive behavioral therapy. Although the risks; (D) there is positive evidence of human fetal risk, but efficacy of improving sleep hygiene recommendations is not potential benefits may warrant use of the drug in pregnant well understood [3], that of cognitive behavioral therapy is women despite potential risks; (X) evidence of human fetal well known to be at least as effective as hypnotic medication, risk, and the risks involved in use of the drug in pregnant and may be longer lasting [24–26]. women clearly outweigh potential benefits. Despite the widespread use of medications to facilitate In Table 2, the generic names of a variety of stimulants sleep, fundamental questions remain largely unanswered, are shown along with the biological half-life, primary mode such as whether the drugs recapitulate normal sleep physi- of excretion, pregnancy class, presence in breast milk, and ology. Uncertainty in this area stems in part from the poor interactions with food, herbs, and smoking tobacco. Lisdex- understanding of which sleep stages (rapid eye movement amfetamine is metabolized rapidly to dextroamphetamine (REM) sleep or non-REM (NREM) sub-stages), or other and the half-life reflects this metabolite’s kinetics (indicated aspects of sleep physiology (such as autonomic or EEG with asterisk). patterns) are most important for the rejuvenating aspects In Table 3, the generic names of miscellaneous sleep- of sleep. To further complicate matters, many drugs used to related drugs are shown along with the biological half-life, induce sleep in patients with insomnia may in fact demon- primary mode of excretion, pregnancy class, presence in strate suppressive effects on aspects of sleep felt by many breast milk, and interactions with food, herbs, and smoking to be fundamentally important, such as slow-wave sleep tobacco. (also known as “deep” sleep or, more formally, as stage N3) In Table 4, major contraindications, beyond allergy to and REM sleep, especially the majority of the mediation or any of its components/metabolites, are as well as traditional benzodiazepines. In a typical night of given for hypnotics, stimulants, and other agents. Note that normal sleep, REM and NREM alternate every approximately coadministration of multiple CNS agents should 90 minutes; the proportion of time spent in N3 tends to always be undertaken with substantial caution. ∗ indicates favor the early part of sleep, while the amount of REM that the only listed contraindication on UpToDate was sleep tends to increase gradually with each cycle through hypersensitivity reaction. Adverse effect profiles are then the night. Beyond these mechanistic questions, important listed according to major clinical categories: neuropsychiatric concerns surround the adverse effect profiles of sleeping pills, systems, Figure 1; cardiovascular system, Figure 2;endo- ranging from “hangover” effect upon waking, to induction crine/metabolic, Figure 3; gastrointestinal and urinary, of parasomnia, to the potential even for increased morbidity Figure 4. Finally, a summary of cytochrome P450 sys- and mortality through a variety of potential mechanisms tem interactions is given in Figure 5. Information contribut- [27]. ing to the text, figures, and tables were assembled from the Here, we review the major clinical considerations in the listed references as well as several review resources [19–21], use of sleeping pills of a variety of classes. ISRN Pharmacology 3

Table 1: Sleeping pill characteristics.

Pregnancy and Generic Half-life Excretion Interactions with food, herbs, smoking lactation Pregnancy class D Smoking and St. John’s Wort may 11.2 h Renal Enters breast milk decrease levels Grapefruit juice may inhibit Pregnancy class C 9–27 h Renal metabolism; St. John’s wort may Enters breast milk decrease levels Pregnancy class C hydrate 8–11 h Hepatic Enters breast milk Pregnancy class D 19–50 h Hepatic St. John’s Wort may decrease levels Enters breast milk Food and grapefruit juice may increase Pregnancy class D 20–50 h Hepatic levels; St. John’s Wort may decrease Enters breast milk levels Pregnancy class B Dihenphydramine 2–10 h Renal Enters breast milk Pregnancy class C High-fat meal increases bioavailability Doxepin 15 h Hepatic Enters breast milk but delays peak level timing Pregnancy class X Grapefruit juice may increase serum 10–24 h Hepatic breast milk: unknown levels Pregnancy class C Eszopiclone 6 h Hepatic Large meals may delay absorption breast milk: unknown Pregnancy unclassified Grapefruit juice may increase serum 2.3 h Hepatic breast milk: unknown levels Pregnancy class D 12.9 h Hepatic Enters breast milk Pregnancy class C Mirtazepine 20–40 h Hepatic St. John’s Wort may decrease levels Enters breast milk Pregnancy class C 2–4 h Hepatic Food decreases availability Enters breast milk Pregnancy unclassified Nortriptyline 28–31 h Renal Enters breast milk Pregnancy unclassified 3–6 h Renal breast milk: unknown Pregnancy class C Food increases absorption rate; St. 6-7 h Hepatic Enters breast milk John’s Wort may decrease levels Pregnancy class C Ramelteon 1–2.6 h Hepatic breast milk: unknown Pregnancy class X Grapefruit juice may increase levels; St. 9–12 h Hepatic Enters breast milk John’s Wort may decrease levels Pregnancy class C Trazodone 7–10 h Hepatic Food increases absorption rate Enters breast milk Pregnancy class X Grapefruit juice may increase levels; St. 1.5–6 h Renal breast milk: unknown John’s Wort may decrease levels Pregnancy class C Zaleplon 1 h Hepatic St. John’s Wort may decrease levels Enters breast milk Food and St. John’s Wort decrease Pregnancy class C Zolpidem 2.5–3 h Hepatic availability; grapefruit juice may Enters breast milk decrease metabolism

2.1. Benzodiazepines. The benzodiazepine family shares these (generally inhibitory) ligand-gated chloride channels, multifunctional clinical effects including anticonvulsant, the clinical impact of these drugs is attributed to neuronal anxiolytic, amnestic, and hypnotic features. All members of inhibition. the benzodiazepine class bind with high affinity to neuronal The benzodiazepines have liabilities in the treatment GABAA receptors, which mediate synaptic and extrasynaptic of insomnia due to concerns for adverse effects, toler- forms of inhibition [28]. By enhancing the conductance of ance, and dependence [29]. The pharmacokinetics of these 4 ISRN Pharmacology Psychiatric sx Memory/cognition Depression Parasomnia Gait and coordination Sensory changes Seizure Suicide Motor changes Dependence/abuse Sleeping/fatigue Insomnia Headache Alprazolam X X X X X X X X X X XXX Amitriptyline X X X X X X X X X X X X X X X X X Clonazepam X X X X X X X X X XXX Diazepam X X X X X X X X X Diphenhydramine X X X X X X X X DoxepinX XXXXXXXX Estazolam X X X X X X X X X Eszopiclone X X X X X X X X X X Flurazepam X X X X X X X X X Lorazepam X X X X X X X X X X Mirtazepine X X X X X X X X

Sleeping pills Nefazodone X X X X X X X X X X Nortriptyline X X X X X X X X X X Oxazepam X X X X X X X Quetiapine X X X X X X X X X X XX Ramelteon X X X X X Temazepam X X X X X X X X Trazodone X X X X X X X X X X Triazolam X X X X X X X X X X Zaleplon X X X X X X X X X X Zolpidem X X X X X X X X X X X Caffeine X X X X Dexmethylphenidate X X X X X X Dextroamphetamine X X X X X Dextroamp/amphet X X X X X X X X

Stimulants Lisdexamfetamine X X X X X X X Methylphenidate X X X X X X X X X X Modafinil X X X X X X X X X X X X X X X X X GHB X X X X X X X X X X XXX

Other Pramipexole X X X X X X X X X X Ropinirole X X X X X X X X X X

X Serious X <1% X 1–10% X >10%

Figure 1: Neuropsychiatric adverse effect profiles. The frequency of reported adverse effects for each drug is given according to the color key. The categories that included a spectrum of symptoms are as follows: parasomnia (abnormal dreams, nightmares, vivid dreams, complex sleep-related behavior, sleep cooking, sleep eating, sleep driving, phone calls, restless legs or leg movements); sleepiness/fatigue (drowsy, hangover, hypersomnia, lethargy, sedation, sluggish, somnolence); gait and coordination (imbalance, dizziness, dysarthria, falls, motion sickness, vertigo); sensory change (hyperesthesia, hypoesthesia, numbness, back pain, myalgia, neuralgia, paresthesia, peripheral neuropathy); motor change (akathisia, ataxia, choreiform movements, dysdiadochokinesis, dyskinesia, hyperkinesias, dystonia, tics, extrapyramidal symptoms, hypertonia, hypotonia, muscle spasm, myoclonus, parkinsonism, Tourette’s, tremor, twitching); memory/cognition (abnormal thinking, attention disturbance, amnesia, cognitive disorder, confusion, mental impairment); psychiatric change (anger, anxiety, apathy, dysphoria, emotional lability, irritability, malaise, mood swings, aggression, agitation, disinhibition, hostility, inappropriate behavior, delirium, delusions, depersonalization, derealization, psychosis, hallucinations, mania, nervousness, tension, paranoia, depression). ISRN Pharmacology 5

Table 2: Stimulant characteristics. Interactions with food, herbs, Generic Half life Excretion Pregnancy and lactation smoking Pregnancy class C Caffeine 5hr Hepatic Enters breast bilk Pregnancy class C High-fat foods may increase Dexmethylphenidate 2–4.5 hr Hepatic breast milk: unknown absorption rate Pregnancy class C Absorption may be altered by acidic Dextroamphetamine 10–13 hr Renal Enters breast milk foods, juices, and vitamin C Dextroamphetamine Pregnancy class C Absorption may be altered by acidic 10–13 hr Renal and amphetamine Enters breast milk foods, juices, and vitamin C

∗ Pregnancy class C High-fat foods may increase Lisdexamfetamine 10–13 hr Hepatic Enters breast milk absorption rate Pregnancy class C Food delays absorption; high-fat Methylphenidate 3-4 hr Renal Enters breast milk meals increase Metadate AUC Pregnancy class C Modafinil 15 hr Hepatic breast milk: unknown

Table 3: Miscellaneous sleep medication characteristics.

Generic Half life Excretion Pregnancy and lactation Interactions with food, herbs, smoking Pregnancy class C Gabapentin 5–7 h Renal Enters breast milk Pregnancy class B GHB 30–60 min n/a High-fat meals decrease peak serum level breast milk: unknown Pregnancy class C Pramipexole 8.5 h Renal breast milk: unknown Pregnancy class C Ropinirole 6 h Hepatic breast milk: unknown medications are variable and should be considered in the reason they are sometimes referred to as nonbenzodiazepine clinical context as each relates to the timing of dosing hypnotics. An important feature of these medications is and the risk of hangover effects, especially the potential the subtype selectivity at the GABAA receptor complex. For risk of sedation carrying into the waking day and affecting example, zolpidem has α1 selectivity, while eszopicole also operation of automobiles or other activities [30]. Other interacts with the α2andα3 subtypes. Transgenic rodent important considerations include the clinical context, such studies have suggested that certain α subtypes mediate cer- as comorbid psychiatric disorders (e.g., cross-tolerance may tain clinical effects of the benzodiazepines (which are fairly occur; also, mood changes can occur), and comorbid nonselective for α1, α2, α3, and α5 subtypes) [36]. abuse advanced age should prompt caution, as paradoxical The effects on sleep architecture are subtle and of effects may be observed, and certain side effect issues become unknown clinical importance. The major clinical concerns particularly important such as gait instability and falls, as surrounding this class include the risk of parasomnia. well as concern for cognitive impairment. These activities may include complex and potentially risky Benzodiazepines tend to suppress slow-wave sleep (N3) behaviors such as eating and even leaving the home or and REM sleep, in favor of increased time spent in stage operating machinery/automobile, without memory of these N2. In addition, the sleep EEG may show increased spindle events. Also worth noting is the pharmacokinetics of these activity and beta frequency. The clinical importance of agents. Zaleplon has the shortest half-life, and is most useful these observations remains poorly understood, although it is for isolated sleep-onset insomnia and may also be considered intriguing that N3 and REM suppression might be expected for awakenings within the night that are too close to final to impair the restorative value of sleep, as these stages have wake time to allow for middle-of-the-night dosing with other received the most attention in the literature as important for agents (for risk of sedation intruding into the waking day). memory and other functions [31–33]. 3. Antidepressants 2.2. Benzodiazepine Site Ligands. This class of new genera- tion hypnotics includes zolpidem, eszopicole, and zaleplon The tricyclic and heterocyclic medications [34, 35]. They differ from traditional benzodiazepines in are often associated with sleepiness, probably related to their chemical structure, but they interact with the ben- antihistamine and possibly other “off-target” receptor inter- zodiazepine binding site on GABAA receptors. For this actions (such as anticholinergic and anti-histamine effects). 6 ISRN Pharmacology

Table 4: Significant contraindications.

Ketoconazole, itraconazole, narrow-angle glaucoma, significant sleep apnea, Alprazolam myasthenia gravis, respiratory insufficiency, liver failure, and pregnancy Amitriptyline Recent or concurrent MAOI use and acute myocardial infarction Chloral hydrate Hepatic or renal impairment, gastritis or ulcer, and severe cardiac disease Narrow-angle glaucoma, significant sleep apnea, myasthenia gravis, respiratory Clonazepam insufficiency, liver failure, and pregnancy Narrow-angle glaucoma, significant sleep apnea, myasthenia gravis, respiratory Diazepam insufficiency, liver failure, and pregnancy Dihenhydramine Acute asthma and breast-feeding Doxepin Recent or concurrent MAOI use, narrow angle glaucoma, and urinary retention Narrow-angle glaucoma, significant sleep apnea, myasthenia gravis, respiratory Estazolam insufficiency, liver failure, and pregnancy Eszopiclone ∗ Narrow-angle glaucoma, significant sleep apnea, myasthenia gravis, respiratory Flurazepam insufficiency, liver failure, and pregnancy Narrow-angle glaucoma, significant sleep apnea, myasthenia gravis, respiratory Lorazepam insufficiency, liver failure, and pregnancy Mirtazepine Recent or concurrent MAOI use Active liver disease, recent or concurrent MAOI use, acute myocardial infarction, Nefazodone concurrent use of , cisapride, or pimozide Nortriptyline Recent or concurrent MAOI use and acute myocardial infarction Narrow-angle glaucoma, significant sleep apnea, myasthenia gravis, respiratory Oxazepam insufficiency, liver failure, and pregnancy Quetiapine ∗ Ramelteon Concurrent fluvoxamine Narrow-angle glaucoma, significant sleep apnea, myasthenia gravis, respiratory Temazepam insufficiency, liver failure, and pregnancy Trazodone ∗ Ketoconazole, itraconazole, narrow-angle glaucoma, significant sleep apnea, Triazolam myasthenia gravis, respiratory insufficiency, liver failure, and pregnancy Zaleplon ∗ OSA, repiratory impairment, myasthenia gravis, severe hepatic impairment, and Zolpidem sleepwalking Caffeine ∗ Agitation, anxiety, glaucoma, motor tics, history of Tourette’s syndrome, and Dexmethylphenidate recent or concurrent MAOI use Symptomatic cardiovascular/atherosclerotic disease, moderate or severe Dextroamphetamine hypertension, hyperthyroidism, glaucoma, agitation, history of drug abuse, and recent or concurrent MAOI use Symptomatic cardiovascular/atherosclerotic disease, moderate or severe Dextroamphetamine and amphetamine hypertension, hyperthyroidism, glaucoma, agitation, history of drug abuse, and recent or concurrent MAOI use Symptomatic cardiovascular/atherosclerotic disease, moderate or severe Lisdexamfetamine hypertension, hyperthyroidism, glaucoma, agitation, history of drug abuse, and recent or concurrent MAOI use Agitation, anxiety, glaucoma, motor tics, history of Tourette’s syndrome, and Methylphenidate recent or concurrent MAOI use Modafinil ∗ Gabapentin ∗ GHB Alcohol and other CNS Pramipexole ∗ Ropinirole ∗ ISRN Pharmacology 7 Hypertension Palpitations Arrythmia Chest discomfort Hypotension Heart attack Alprazolam X X X Amitriptyline X X X X X Chloral hydrate Clonazepam X Diazepam X Diphenhydramine X X X X Doxepin X X X X X Estazolam X Eszopiclone X X Flurazepam X X X Lorazepam X Mirtazepine X X

Sleeping pills Nefazodone X X X Nortriptyline X X X X X Oxazepam Quetiapine X X X X X Ramelteon Temazepam Trazodone X X X X X X Triazolam X Zaleplon X X X X X Zolpidem X X X X Caffeine X X X Dexmethylphenidate Dextroamphetamine X X Dextroamp/amphet X X

Stimulants Lisdexamfetamine X Methylphenidate X X X X X Modafinil X X X Gabapentin GHB X X X X

Other Pramipexole X X RopiniroleXXXXXX

X Serious X <1% X 1–10% X >10%

Figure 2: Cardiovascular adverse effect profiles. The frequency of reported adverse effects for each drug is given according to the color key. The categories that included a spectrum of symptoms are as follows: arrhythmia (atrial, ventricular, A-V block, bundle branch block, extrasystole); hypotension (orthostatic).

Most neuroactive drugs exhibit what has become known acetylcholine receptors, and histamine H1 receptors. Other as “promiscuity”, that is, interaction with multiple targets, traditional antidepressants used as sleeping pills include although the clinical implications remain to be fully explored trazodone, nefadozone, and mirtazepine. Trazodone is a 5HT [37–41]. Common tricyclics used as sleeping pills include reuptake inhibitor with antagonism also at 5HT1A,5HT2A, amitriptyline, nortriptyline (which is itself an active metabo- and α1 adrenergic receptors, with low affinity interaction at lite of amitriptyline), and low doses of doxepin. These agents histamine H1 receptors. Priapism is a rare but serious side have high affinity for the 5HT and NE reuptake enzymes as effect of trazodone. Mirtazepine is an antagonist at multiple well as 5HT2A receptors, α1 adrenergic receptors, muscarinic subtypes of the 5HT2 receptor, as well as the α2 adrenergic 8 ISRN Pharmacology Thyroid changes Weight gain Weight loss Blood sugar changes Appetite changes Menstrual changes Sexual dysfunction Alprazolam X X X X X Amitriptyline X Chloral hydrate Clonazepam X X X X X Diazepam X Diphenhydramine X X Doxepin X X X X Estazolam X X Eszopiclone X X X Flurazepam X X X Lorazepam X X X X Mirtazepine X X X

Sleeping pills Nefazodone X X Nortriptyline X X X X Oxazepam XX Quetiapine X X X X X Ramelteon Temazepam X Trazodone X X X X X Triazolam Zaleplon X X X X X Zolpidem X X X Caffeine Dexmethylphenidate X Dextroamphetamine X X X Dextroamp/amphet X X X X Lisdexamfetamine X X X Stimulants Methylphenidate X X X X Modafinil X Gabapentin X X X X GHB X X X

Other Pramipexole X X X Ropinirole X X X X

X Serious X <1% X 1–10% X >10%

Figure 3: Endocrine/metabolic adverse effect profiles. The frequency of reported adverse effects for each drug is given according to the color key. The categories that included a spectrum of symptoms are as follows: thyroid (hyper- or hypothyroidism), blood sugar (diabetes, increased or decreased blood sugar); appetite change (increased or decreased, dysphagia); menstrual change (amenorrhea, dysmenorrheal, early menses, menorrhagia, menstrual cramps, menstrual disorders); sexual dysfunction (erectile dysfunction, delayed ejaculation, impotence, change in libido, priapism).

receptor, and the histamine H1 receptor. Nefazodone is In contrast to the above antidepressant agents, the newer also an antagonist of 5HT2 receptors, α2 adrenergic recep- SSRI and SNRI medications are generally either neutral from tors, and the histamine H1 receptors. Hepatotoxicity is a an alertness standpoint or are activating and thus unlikely prominent clinical concern for nefazodone. Kidney and liver to be prescribed for the purpose of insomnia treatment, function may influence the choice of agents in this class. except in the indirect setting when the insomnia is felt to ISRN Pharmacology 9 Abnormal taste Nausea/vomiting Acid reflux Liver enzyme elevation Abdominal pain Urinary symptoms Kidney function changes Diarrhea Constipation Alprazolam X X X X X Amitriptyline X X X X X Chloral Hydrate X X Clonazepam X X X X X X Diazepam X X X X Diphenhydramine X X X X Doxepin X X X X X X X X Estazolam X X X X Eszopiclone X X X X X Flurazepam X X X X X X Lorazepam X Mirtazepine X X X X X

Sleeping pills Nefazodone X X X X X X X Nortriptyline X X X X X X X Oxazepam X Quetiapine X X X X X X X X Ramelteon X X Temazepam X X Trazodone X X X X X X X Triazolam X Zaleplon X X X X X X X Zolpidem X X X X X X X X Caffeine X X Dexmethylphenidate X X Dextroamphetamine X X X Dextroamp/amphet X X X X X Lisdexamfetamine X X X Stimulants Methylphenidate X X X X X X Modafinil X X X X X X Gabapentin X X X X X GHB X X X X X X X X X

Other Pramipexole X X X X X Ropinirole X X X X X X X

X Serious X <1% X 1–10% X >10%

Figure 4: Gastrointestinal and genitor-urinary adverse effect profiles. The frequency of reported adverse effects for each drug is given according to the color key. The categories that included a spectrum of symptoms are as follows: change in liver enzymes (transaminase, bilirubin, hepatic failure, hepatic encephalopathy or coma, hepatitis); urinary symptoms (abnormal urine, enuresis, dysuria, pyuria, hematuria, difficulty urinating, nocturia, polyuria, incontinence, retention, change in frequency); change in kidney function (hyponatremia, acute renal failure).

be a consequence of a mood disorder. These drugs may also 3.1. Anticonvulsants. Certain anticonvulsants confer seda- increase motor activity during sleep; whether shifting daily tion as an adverse effect, and may provide some potential dosing to the morning improves these nocturnal side effects benefit for insomnia complaints, especially if there is already is patient-specific. an indication for the medication. For example, gabapentin 10 ISRN Pharmacology

1A2 2A6 2B6 2C8 2C9 2C19 2D6 2E1 3A4 Alprazolam S Amitriptyline S S S S S S Chloral hydrate S Clonazepam S Diazepam S S S S S Diphenhydramine Doxepin S S S S Estazolam S Eszopiclone S S Flurazepam S Lorazepam∗ Mirtazepine S S SS

Sleeping pills Nefazodone SS Nortriptyline S S S S Oxazepam∗ Quetiapine S S Ramelteon S S S Temazepam S S S S Trazodone S S Triazolam S Zaleplon S Zolpidem S S S S S Caffeine S SSSS Dextroamphetamine ? ? ? Dextroamp/amphet Dexmethylphenidate

Stimulants Lisdexamfetamine ? ? ? Methylphenidate Modafinil S Gabapentin GHB

Other Pramipexole Ropinirole S S

Minor inhib Moderate inhib Major inhibition S: minor substr S: major substr

Figure 5: Cytochrome P450 profiles. The metabolism and interactions with the CYP family enzymes is given for each drug. The features of the interaction are given in the color key. ? refers to the substrate interactions being limited to in vitro studies.

may be used for certain kinds of pain, and could have preferring, but also shows antagonism of 5HT2 receptors beneficial effects for symptomatic relief of RLS and/or and histamine H1 receptors. The metabolite norquetiapine insomnia. Gabapentin is thought to act on neuronal calcium inhibits NE reuptake. channels, but the mechanism(s) of sedation remain uncer- tain. Peripheral edema and neuropsychiatric side effects 3.3. Melatonin System. Melatonin is released by the pineal remain prominent concerns for gabapentin. gland through regulation from the suprachiasmatic nucleus, mainly driven by light exposure conveyed by the retinohy- 3.2. Neuroleptics. The dopamine antagonists generally cause pothalamic tract [42].Tworeceptortypes,theMT1and some degree of sedation, and patients requiring these medi- MT2, medicate the molecular signaling of melatonin. Secre- cations for psychiatric reasons may experience improvement tion follows a circadian rhythm, and exogenous administra- in their comorbid insomnia. Quetiapine may be the most tion can alter a patient’s circadian rhythm depending on the commonly prescribed neuroleptic for the off-label use as a timing of dosing. For example, early evening administration hypnotic. However, this drug has substantial adverse effect of melatonin results in a phase advance, which may be burden (including weight gain, blood sugar handling, etc.), implemented to address delayed circadian phase syndrome with particular concern for older populations. Its binding [15]. Melatonin is available over the counter and has profile is that of atypical neuroleptics in that it is D2 extensive literature supporting its use in circadian rhythm ISRN Pharmacology 11 disorders, although the evidence for its use for insomnia is thus they should be avoided unless other backup forms limited [43]. Ramelteon is a synthetic of contraception are used. Patients should also be aware agonist, which is nonselective for the two subtypes. of potential interactions of prescription stimulants with over-the-counter products that contain stimulants such as 3.4. Antihistamines. The nonselective antihistamines such caffeine, pseudoephedrine, ephedrine. as diphenhydramine may produce sleepiness as the result of inhibition of central histamine H1 receptors. Histamine 4.2. Caffeine. Caffeine is contained in a diversity of food and is one of several wake-promoting neurotransmitters and is drink and is the most widely used stimulant. Individual released by the diffuse projections of the tuberomammillary differences in the response to caffeine, including sensitivity to nucleus [44]. Over-the-counter preparations may include side effects, may be due to polymorphisms in the adenosine combinations of diphenhydramine with analgesics such as signaling system [58–61]. Because caffeine can lead to similar ibuprofen and acetaminophen. The main concerns in the use cardiovascular and CNS effects as the other stimulants, of antihistamines for insomnia are the long half-life (which caution should be exercised in regards to the possibil- tends to produce a hangover-like effect), and in those of ity that patients may augment prescription medications advanced age, the anticholinergic effects may alter cognition. with caffeine-containing products or other over-the-counter At high doses, EKG changes become an increasing concern. agents that may present interaction concerns. Prescription antihistamines have also been used off-label, such as . 5. Sleep-Related Movement Disorders 4. Sleepiness The medications pramipexole and ropinirole were initially developed for treatment of Parkinson’s disease. However, Sleepiness is a common daytime complaint, with a diversity they have been shown to improve RLS symptoms, which are of potential etiologies [45]. Although some promote the also thought to derive from impaired dopaminergic signaling distinction between sleepiness and fatigue as a gateway to in the brain [16, 62]. This may underlie the improvement in considering etiologies, the language patients use to describe RLS that may accompany oral iron repletion, as manifested this spectrum of symptoms is as imprecise as the medical by low serum ferritin, due to iron being a cofactor in the rate- profession’s ability to quantify them [46–49]. For example, limiting step of brain dopamine synthesis. One main concern the commonly employed Epworth sleepiness scale has only with these agents relates to orthostatic hypotension, which meager correlation with sleep apnea severity [50–52]. Even may be of particular interest to those with nocturia due objective measures of sleepiness, such as the sleep latency to syncope/fall risk during nocturnal awakenings. Although values in the multiple sleep latency test (MSLT), show weak very rare, compulsive behaviors (such as gambling) may relation to underlying sleep apnea severity [53]. Adding occur with the dopaminergic agents. Kidney and liver to the complexity of the situation, some patients with function may dictate the choice between these two drugs, effectively treated primary causes of sleepiness, such as as pramipexole undergoes renal clearance while ropinirole sleep apnea, may exhibit persistent or residual sleepiness. undergoes hepatic metabolism. It is critical to determine underlying causes for excessive Clonazepam (with or without melatonin) is commonly sleepiness before considering countermeasures in the form used for REM sleep behavior disorder [18]. RBD is char- of stimulant therapy. For example, stimulant treatments may acterized on the PSG as REM without atonia (and possibly mask symptoms that in fact point to treatable factors such as capturing dream enactment), and by the clinical history of sleep apnea, insufficient sleep, etc. dream enactment, which should be distinguished from other nocturnal behaviors such as NREM parasomnias or seizure 4.1. Stimulants. The largest class of stimulants includes activity or delirium. RBD episodes may include vocalizations amphetamines or amphetamine derivatives, although other as well as actions, which can be associated with injurious options include modafinil and caffeine [54–57]. They are behavior to the patient and/or bed partner. In addition to thought to enhance alertness by increasing the release of pharmacological suppression with clonazepam, it may be catecholamines, known to be excitatory and thus increase useful to assess and treat sleep apnea, which may contribute wakefulness. The pharmacokinetics (half-life) should be con- to REM-related arousals. sidered in treatment strategies, keeping in mind the potential for patient-specific effects. Avoiding consumption later in 6. Metabolic Considerations: the day is an important consideration for the potential of The Cytochrome P450 System inadvertently causing insomnia symptoms. Kidney and liver function may influence the choices in this class. The cytochrome P450 (CYP) family of heme-containing Modafinil and armodafinil are nonamphetamine stimu- enzymes is involved in the metabolism of organic molecules, lants with a mechanism that may involve dopamine signaling both endogenous and exogenous [63–65]. These proteins, among other possibilities [54]. Like the amphetamines, side found mainly in the liver, transform their substrates through effects relate to the activating properties. In addition, the oxidation, an important step in metabolism. Each CYP rare but serious Stevens-Johnson reaction has been reported enzyme may catalyze the metabolism of multiple drugs, with modafinil. For women taking hormonal contraception and a single drug may interact with one or more CYP pills, their levels may be decreased by these agents and enzymes. Before drugs reach the liver, metabolism may 12 ISRN Pharmacology begin in the intestine primarily via the CYP3A4 isozyme. cross-refer CYP considerations shown here with those related We summarize the relevant CYP interactions in regards to to the rest of their medications [63, 64]. It is worth noting the drugs mentioned above; like drug-drug interactions, it that nonprescription drugs and certain food and natural is suggested that medical databases be consulted for more supplement products may influence the CYP system [66]. complete information. For example, certain substances found in grapefruit juice There are two main contexts for considering the CYPs in inhibit the CYP3A4 isozyme. In addition, the natural supple- clinical pharmacology. The most clinically relevant is the ment St. John’s Wort induces CYP3A4 and inhibits CYP1A1, effect of certain drugs on enhancing or inhibiting the action CYP1B1, CYP2D6, and CYP3A4. Smoking tobacco, as well of one or more CYP enzymes, which may alter the meta- as consumption of char-broiled foods, leads to induction of bolism of shared substrate drugs. The other context is the CYP1A2 isozyme [64]. Less common interactions include the increasing recognition that genetic polymorphisms may alter inhibition of CYP2A6 by starfruit, and inhibition of CYP2E1 an individual’s metabolism of certain drugs. For example, by watercress. genetic variants of a CYP enzyme yield phenotypic classifica- tion into poor metabolizer, extensive metabolizer, and ultra- Acknowledgments rapid metabolizer phenotypes [64]. Homozygous individuals for the autosomal recessive allele are poor metabolizers, Dr. Bianchi receives funding from the Department of Neu- and individuals that are heterozygous or or wild type are rology, Massachusetts General Hospital, a Young Clinician extensive metabolizers. Duplication or amplification results Award from the Center for Integration of Medicine and in the ultrarapid metabolizer phenotype. These phenotypes Innovative Technology, and a Harvard Catalyst KL2 Medical may have important implications for drug development and Research Investigator Fellowship. clinical trial design as well as clinical care [63, 64]. The impact of drug interactions related to the CYP system range References from the need for dose adjustments to the risk of important toxicities. Age is another consideration, as advanced age may [1] E. Senthilvel, D. Auckley, and J. Dasarathy, “Evaluation of sleep be associated with reduced CYP activity [65]. Liver and heart disorders in the primary care setting: history taking compared failure may affect the CYP system as well. to questionnaires,” Journal of Clinical Sleep Medicine, vol. 7, Inhibition of a CYP enzyme may be reversible or irre- no. 1, pp. 41–48, 2011. versible. For reversible inhibition, if two drugs are substrates [2] J. Winkelman and R. Pies, “Current patterns and future for a CYP enzyme, the drug with the more potent interaction directions in the treatment of insomnia,” Annals of Clinical will undergo more extensive metabolism, while the less Psychiatry, vol. 17, no. 1, pp. 31–40, 2005. [3] M. K. Erman, “Therapeutic options in the treatment of inso- potent CYP substrates will undergo less metabolism. This mnia,” Journal of Clinical Psychiatry, vol. 66, no. 9, pp. 18–23, form of competitive CYP inhibition will increase serum levels 2005. of the less potent drug, which may have important toxicity [4] T. J. Balkin, T. Rupp, D. Picchioni, and N. J. Wesensten, “Sleep implications. For prodrugs that require CYP processing to loss and sleepiness: current issues,” Chest, vol. 134, no. 3, pp. generate an active metabolite, CYP inhibition will actually 653–660, 2008. reduce the amount of the active drug. Irreversible inhibition [5] M. M. Ohayon, “From wakefulness to excessive sleepiness: is less common, and may require increased time after drug what we know and still need to know,” Sleep Medicine Reviews, discontinuation before the CYP enzyme activity is restored vol. 12, no. 2, pp. 129–141, 2008. to baseline. [6] International Classification of Sleep Disorders, Diagnostic CYP enzyme induction can occur via increased syn- & Coding Manual, American Academy of Sleep Medicine, thesis of the enzyme, or increase in the rate of enzymatic Westchester, Ill, USA, 2nd edition, 2005. [7] A.D.Krystal,“Theeffect of insomnia definitions, terminology, metabolism of existing proteins. CYP1A2, CYP2C9, CYP2E1, and classifications on clinical practice,” Journal of the American and CYP3A4 are inducible. Induction may be gradual if Geriatrics Society, vol. 53, no. 7, pp. S258–S263, 2005. synthesis is involved, unlike inhibition which tends to be [8] J. D. Edinger, M. H. Bonnet, R. R. Bootzin et al., “Derivation more acute. of research diagnostic criteria for insomnia: report of an The CYP metabolism of sleep-wake drugs is shown American Academy of Sleep Medicine work group,” Sleep, vol. in Figure 5. In each case, the substrate or inhibitor status 27, no. 8, pp. 1567–1596, 2004. is indicated, along with whether this is major or minor [9] D. E. Moul, E. A. Nofzinger, P. A. Pilkonis, P. R. Houck, J. M. (the former obviously requiring closer consideration). Some Miewald, and D. J. Buysse, “Symptom reports in severe chronic drugs are both substrates for, and inhibitors of, a single insomnia,” Sleep, vol. 25, no. 5, pp. 553–563, 2002. CYP enzyme (e.g., nefazodone with 3A4). For example, most [10] M. M. Ohayon, “Epidemiology of insomnia: what we know sleeping pills are substrates for CYP 3A4, which is involved and what we still need to learn,” Sleep Medicine Reviews, vol. 6, in the metabolism of many common drugs (e.g., the statins no. 2, pp. 97–111, 2002. [11] W. V. McCall, “A psychiatric perspective on insomnia,” Journal atorvastatin and simvastatin), a reminder that combination of Clinical Psychiatry, vol. 62, no. 10, pp. 27–32, 2001. therapy may require adjustment of dose or reassessment [12] A. G. Harvey, “Insomnia: symptom or diagnosis?” Clinical of risk-benefit balance. Another example is modafinil and Psychology Review, vol. 21, no. 7, pp. 1037–1059, 2001. 2C19, which also metabolizes proton pump inhibitors. [13] M. J. Sateia, K. Doghramji, P. J. Hauri, and C. M. Morin, Because many patients with sleep disorders also have comor- “Evaluation of chronic insomnia,” Sleep,vol.23,no.2,pp. bid medical and psychiatric disorders, it is important to 243–208, 2000. ISRN Pharmacology 13

[14] A. G. Harvey and N. K. Tang, “(Mis)perception of sleep [35] D. J. Nutt and S. M. Stahl, “Searching for perfect sleep: the in insomnia: a puzzle and a resolution,” Psychological Bulletin, continuing evolution of GABAA receptor modulators as vol. 138, no. 1, pp. 77–101, 2012. hypnotics,” Journal of Psychopharmacology, vol. 24, no. 11, pp. [15] T. I. Morgenthaler, T. Lee-Chiong, C. Alessi et al., “Practice 1601–1612, 2010. parameters for the clinical evaluation and treatment of circa- [36] M. T. Bianchi, “The pharmacology of insomnia: targeting dianrhythmsleepdisorders:anAmericanAcademyofSleep GABAA receptor function,” International Journal of Sleep Medicine report,” Sleep, vol. 30, no. 11, pp. 1445–1459, 2007. Disorders, vol. 1, pp. 102–110, 2007. [16] M. Zucconi and L. Ferini-Strambi, “Epidemiology and clinical [37] M. T. Bianchi and E. J. Botzolakis, “Targeting ligand-gated findings of restless legs syndrome,” Sleep Medicine, vol. 5, no. ion channels in neurology and psychiatry: is pharmacological 3, pp. 293–299, 2004. promiscuity an obstacle or an opportunity?” BMC Pharmacol- [17] J. Haba-Rubio, L. Staner, J. Krieger, and J. P. MacHer, “What ogy, vol. 10, article 3, 2010. is the clinical significance of periodic limb movements during [38] M. T. Bianchi, “Promiscuous modulation of ion channels sleep?” Neurophysiologie Clinique, vol. 34, no. 6, pp. 293–300, by anti-psychotic and anti-dementia medications,” Medical 2004. Hypotheses, vol. 74, no. 2, pp. 297–300, 2010. [18] J. F. Gagnon, R. B. Postuma, S. Mazza, J. Doyon, and J. [39] M. T. Bianchi, J. Pathmanathan, and S. S. Cash, “From ion Montplaisir, “Rapid-eye-movement sleep behaviour disorder channels to complex networks: magic bullet versus magic and neurodegenerative diseases,” Lancet Neurology, vol. 5, no. shotgun approaches to anticonvulsant pharmacotherapy,” 5, pp. 424–432, 2006. Medical Hypotheses, vol. 72, no. 3, pp. 297–305, 2009. [19] http://www.uptodate.com/index. [40] M. T. Bianchi, “Non-serotonin anti-depressant actions: direct [20] http://www.thompsonhc.com/micromedex2/. ion channel modulation by SSRIs and the concept of single [21] T. Barkoukis, J. K. Matheson, R. Ferber, and K. Doghramji, agent poly-pharmacy,” Medical Hypotheses,vol.70,no.5,pp. Therapy in Sleep Medicine, Elsevier, 2012. 951–956, 2008. [22] R. M. Benca, “Diagnosis and treatment of chronic insomnia: a [41] G. Rammes and R. Rupprecht, “Modulation of ligand-gated review,” Psychiatric Services, vol. 56, no. 3, pp. 332–343, 2005. ion channels by antidepressants and ,” Molecular [23] A. L. Meolie, C. Rosen, D. Kristo et al., “Oral nonprescription Neurobiology, vol. 35, no. 2, pp. 160–174, 2007. treatment for insomnia: an evaluation of products with limit- [42] F. W. Turek and M. U. Gillette, “Melatonin, sleep, and ed evidence,” Journal of Clinical Sleep Medicine, vol. 1, no. 2, circadian rhythms: rationale for development of specific pp. 173–187, 2005. melatonin agonists,” Sleep Medicine, vol. 5, no. 6, pp. 523–532, [24] L. Kierlin, “Sleeping without a pill: nonpharmacologic treat- 2004. ments for insomnia,” Journal of Psychiatric Practice, vol. 14, no. [43] N. Buscemi, B. Vandermeer, N. Hooton et al., “The efficacy 6, pp. 403–407, 2008. and safety of exogenous melatonin for primary sleep disorders: [25] T. Morgenthaler, M. Kramer, C. Alessi et al., “Practice a meta-analysis,” Journal of General Internal Medicine, vol. 20, parameters for the psychological and behavioral treatment no. 12, pp. 1151–1158, 2005. of insomnia: an update. An American Academy of Sleep [44] C. B. Saper, G. Cano, and T. E. Scammell, “Homeostatic, Medicine Report,” Sleep, vol. 29, no. 11, pp. 1415–1419, 2006. circadian, and emotional regulation of sleep,” Journal of Com- [26] J. D. Edinger and M. K. Means, “Cognitive-behavioral therapy parative Neurology, vol. 493, no. 1, pp. 92–98, 2005. for primary insomnia,” Clinical Psychology Review, vol. 25, no. [45] M. J. Thorpy, “Which clinical conditions are responsible for 5, pp. 539–558, 2005. impaired alertness?” Sleep Medicine, vol. 6, supplement 1, pp. [27] D. F. Kripke, “Chronic hypnotic use: deadly risks, doubtful S13–S20, 2005. benefit,” Sleep Medicine Reviews, vol. 4, no. 1, pp. 5–20, 2000. [46] A. Shahid, J. Shen, and C. M. Shapiro, “Measurements of [28] H. Mohler, “GABA(A) receptor diversity and pharmacology,” sleepiness and fatigue,” Journal of Psychosomatic Research, vol. Cell and Tissue Research, vol. 326, pp. 505–516, 2006. 69, no. 1, pp. 81–89, 2010. [29] A. M. Holbrook, R. Crowther, A. Lotter, C. Cheng, and D. [47] S. S. Sullivan and C. A. Kushida, “Multiple sleep latency test King, “Meta-analysis of benzodiazepine use in the treatment and maintenance of wakefulness test,” Chest, vol. 134, no. 4, of insomnia,” Canadian Medical Association Journal, vol. 162, pp. 854–861, 2008. no. 2, pp. 225–233, 2000. [48] J. Shen, J. Barbera, and C. M. Shapiro, “Distinguishing sleepi- [30] S. R. Pandi-Perumal, J. C. Verster, L. Kayumov et al., “Sleep ness and fatigue: focus on definition and measurement,” Sleep disorders, sleepiness and traffic safety: a public health men- Medicine Reviews, vol. 10, no. 1, pp. 63–76, 2006. ace,” Brazilian Journal of Medical and Biological Research, vol. [49] G. Curcio, M. Casagrande, and M. Bertini, “Sleepiness: 39, no. 7, pp. 863–871, 2006. evaluating and quantifying methods,” International Journal of [31] G. Tononi, “The neuro-biomolecular basis of alertness in sleep Psychophysiology, vol. 41, no. 3, pp. 251–263, 2001. disorders,” Sleep Medicine, vol. 6, no. 1, pp. S8–S12, 2005. [50] R. D. Chervin and M. S. Aldrich, “The multiple sleep latency [32] A. Vassalli and D. J. Dijk, “Sleep function: current questions test and Epworth sleepiness scale in the assessment of daytime and new approaches,” European Journal of Neuroscience, vol. sleepiness,” Journal of Sleep Research, vol. 9, no. 4, pp. 399–401, 29, no. 9, pp. 1830–1841, 2009. 2000. [33] T. T. Dang-Vu, M. Desseilles, P. Peigneux, and P. Maquet, “A [51] N. A. Eiseman, B. B. Westover, J. E. Mietus, R. J. Thomas, role for sleep in brain plasticity,” Pediatric Rehabilitation, vol. and M. T. Bianchi, “Classification algorithms for predicting 9, no. 2, pp. 98–118, 2006. sleepiness and sleep apnea severity,” Journal of Sleep Research, [34] Y. Dundar,¨ S. Dodd, J. Strobl, A. Boland, R. Dickson, and T. vol. 21, no. 1, pp. 101–112, 2012. Walley, “Comparative efficacy of newer hypnotic drugs for the [52] D. J. Gottlieb, C. W. Whitney, W. H. Bonekat et al., “Relation short-term management of insomnia: a systematic review and of sleepiness to respiratory disturbance index: the sleep heart meta-analysis,” Human Psychopharmacology,vol.19,no.5,pp. health study,” American Journal of Respiratory and Critical Care 305–322, 2004. Medicine, vol. 159, no. 2, pp. 502–507, 1999. 14 ISRN Pharmacology

[53] R. D. Chervin, M. S. Aldrich, R. Pickett, and C. Guilleminault, “Comparison of the results of the Epworth Sleepiness Scale and the Multiple Sleep Latency Test,” Journal of Psychosomatic Research, vol. 42, no. 2, pp. 145–155, 1997. [54] J. S. Ballon and D. Feifel, “A systematic review of modafinil: potential clinical uses and mechanisms of action,” Journal of Clinical Psychiatry, vol. 67, no. 4, pp. 554–566, 2006. [55] M. H. Bonnet, T. J. Balkin, D. F. Dinges, T. Roehrs, N. L. Rogers, and N. J. Wesensten, “The use of stimulants to modify performance during sleep loss: a review by the Sleep Deprivation and Stimulant Task Force of the American Academy of Sleep Medicine,” Sleep, vol. 28, no. 9, pp. 1163– 1187, 2005. [56] B. Boutrel and G. F. Koob, “What keeps us awake: the neu- ropharmacology of stimulants and wakefulness-promoting medications,” Sleep, vol. 27, no. 6, pp. 1181–1194, 2004. [57] A. Nehlig, “Are we dependent upon coffee and caffeine? A review on human and animal data,” Neuroscience and Biobe- havioral Reviews, vol. 23, no. 4, pp. 563–576, 1999. [58] E. Childs, C. Hohoff,J.Deckert,K.Xu,J.Badner,andH.de Wit, “Association between ADORA2A and DRD2 polymor- phisms and caffeine-induced anxiety,” Neuropsychopharmacol- ogy, vol. 33, no. 12, pp. 2791–2800, 2008. [59] J. V. Retey,´ M. Adam, E. Honegger et al., “A functional genetic variation of adenosine deaminase affects the duration and intensity of deep sleep in humans,” Proceedings of the National Academy of Sciences of the United States of America, vol. 102, no. 43, pp. 15676–15681, 2005. [60] B. B. Fredholm, K. Battig,¨ J. Holmen,´ A. Nehlig, and E. E. Zvar- tau, “Actions of caffeine in the brain with special reference to factors that contribute to its widespread use,” Pharmacological Reviews, vol. 51, no. 1, pp. 83–133, 1999. [61] T. V. Dunwiddie and S. A. Masino, “The role and regulation of adenosine in the central nervous system,” Annual Review of Neuroscience, vol. 24, pp. 31–55, 2001. [62] M. R. Littner, C. Kushida, W. M. Anderson et al., “Practice parameters for the dopaminergic treatment of restless legs syndrome and periodic limb movement disorder,” Sleep, vol. 27, no. 3, pp. 557–559, 2004. [63] T. Lynch and A. Price, “The effect of cytochrome P450 metabolism on drug response, interactions, and adverse effects,” American Family Physician, vol. 76, no. 3, pp. 391–396, 2007. [64] R. Sikka, B. Magauran, A. Ulrich, and M. Shannon, “Bench to Bedside: pharmacogenomics, adverse drug interactions, and the cytochrome P450 system,” Academic Emergency Medicine, vol. 12, no. 12, pp. 1227–1235, 2005. [65] C. C. Ogu and J. L. Maxa, “Drug interactions due to cyto- chrome P450,” Proceedings (Baylor University. Medical Center), vol. 13, pp. 421–423, 2000. [66] A. A. Izzo and E. Ernst, “Interactions between herbal medi- cines and prescribed drugs: an updated systematic review,” Drugs, vol. 69, no. 13, pp. 1777–1798, 2009.

Question 2

Relevant medical or scientific evidence pertaining to the disease or condition

Contents

Overview – 3

Insomnia, American Academy of Sleep Medicine – 4

Overview

Insomnia is a sleep disorder that can have a significant negative impact on a patient’s daily life. Symptoms include:

• Fatigue • Moodiness • Irritability or anger • Daytime sleepiness • Anxiety about sleep • Lack of concentration • Poor Memory • Poor quality performance at school or work • Lack of motivation or energy • Headaches or tension • Upset stomach • Mistakes/accidents at work or while driving

Treatments include:

• Cognitive behavioral therapy (CBT) • Over-the-counter products • Prescription sleeping pills • Unapproved prescription drugs

Insomnia is a difficult condition to treat due to its variable causes. As a result, many treatments are offered including unapproved prescription drugs that carry serious side effects. A paper by the American Academy of Sleep Medicine provides more information on this topic below.

Insomnia

Insomnia is a common sleep complaint that occurs when you have one or more of these problems: • You have a hard time initiating sleep. • You struggle to maintain sleep, waking up frequently during the night. • You tend to wake up too early and are unable to go back to sleep. • You sleep is nonrestorative or of poor quality.

These symptoms of insomnia can be caused by a variety of biological, psychological and social factors. They most often result in an inadequate amount of sleep, even though the sufferer has the opportunity to get a full night of sleep. Insomnia is different from sleep deprivation, which occurs when an individual does not have the opportunity to get a full night of sleep. A small percentage of people who have trouble sleeping are actually short sleepers who can function normally on only five hours of sleep or less.

There are two types of insomnia – primary and secondary. Primary insomnia is sleeplessness that cannot be attributed to an existing medial, psychiatric or environmental cause (such as drug abuse or medications). Secondary insomnia is when symptoms of insomnia arise from a primary medical illness, mental disorders or other sleep disorders. It may also arise from the use, abuse or exposure to certain substances.

Prevalence

• About 30 percent of adults have symptoms of insomnia • About 10 percent of adults have insomnia that is severe enough to cause daytime consequences • Less than 10 percent of adults are likely to have chronic insomnia

Types

Insomnia is considered a disorder only when it causes a significant amount of distress or anxiety, or when it results in daytime impairment. The International Classification nd of Sleep Disorders, 2 Edition, documents the following types of insomnia:

• Adjustment insomnia: This is also called acute insomnia or short-term insomnia. It is usually caused by a source of stress and tends to last for only a few days or weeks. Epidemiologic studies indicate that the one-year prevalence of adjustment insomnia among adults is likely to be in the range of 15-20%. Adjustment insomnia can occur at any age, although establishing a relationship between a particular stress and sleep disturbance may be difficult in infants. Adjustment insomnia is more common in women than men and in older adults than younger adults and children

• Behavioral insomnia of childhood: Two primary types of insomnia affect children. Sleep-onset association type occurs when a child associates falling asleep with an action (being held or rocked), object (bottle) or setting (parents’ bed), and is unable to fall asleep if separated from that association. Limit-setting type occurs when a child stalls and refuses to go to sleep in the absence of strictly enforced bedtime limits. Approximately 10-30% of children are affected by this condition

• Idiopathic insomnia: An insomnia that begins in childhood and is lifelong, it cannot be explained by other causes. Information suggests that this condition is present in approximately .7% of adolescents and 1.0% of very young adults

• Inadequate sleep hygiene: This form of insomnia is caused by bad sleep habits that keep you awake or bring disorder to your sleep schedule. This condition is present in 1-2% of adolescents and young adults. This condition affected 5-10% of sleep-clinic populations.

• Insomnia due to drug or substance, medical condition, or mental disorder: Symptoms of insomnia often result from one of these causes. Insomnia is associated more often with a psychiatric disorder, such as depression, than with any other medical condition. Surveys suggest approximately 3% of the population has insomnia symptoms that are caused by a medical or psychiatric condition. Among adolescents and young adults, the prevalence of this form of insomnia is slightly lower. 2% of the general population is affected by this type of insomnia. Approximately 3.5% of all sleep-center patients are affected by this condition.

• Paradoxical insomnia: A complaint of severe insomnia occurs even though there is no objective evidence of a sleep disturbance. The prevalence in the general population is not known. Among clinical populations, this condition is typically found in less than 5% of patients with insomnia. It is thought to be most common in young and middle-aged adults.

• Psychophysiological insomnia: A complaint of insomnia occurs along with an excessive amount of anxiety and worry regarding sleep and sleeplessness. This condition is found in 1-2% of the general population and 12-15% of all patients seen at sleep centers. It is more frequent in women than in men. It rarely occurs in young children but is more common in adolescents and all adult age groups

Risk Groups

• A high rate of insomnia is seen in middle-aged and older adults. Although your individual sleep need does not change as you age, physical problems can make it more difficult to sleep well. • Women are more likely than men to develop insomnia. • People who have a medical or psychiatric illness, including depression, are at risk for insomnia. • People who use medications may experience insomnia as a side-effect.

Effects

• Fatigue • Moodiness • Irritability or anger • Daytime sleepiness • Anxiety about sleep • Lack of concentration • Poor Memory • Poor quality performance at school or work • Lack of motivation or energy • Headaches or tension • Upset stomach • Mistakes/accidents at work or while driving

Severe daytime sleepiness typically is an effect of sleep deprivation and is less common with insomnia. People with insomnia often underestimate the amount of sleep they get each night. They worry that their inability to sleep will affect their health and keep them from functioning well during the day. Often, however, they are able to perform well during the day despite feeling tired.

Treatments

• Cognitive behavioral therapy (CBT): CBT can have beneficial effects that last well beyond the end of treatment. It involves combinations of the following therapies: o Cognitive therapy: Changing attitudes and beliefs that hinder your sleep o Relaxation training: Relaxing your mind and body o Sleep hygiene training: Correcting bad habits that contribute to poor sleep o Sleep restriction: Severely limiting and then gradually increasing your time in bed o Stimulus control: Going to bed only when sleepy, waking at the same time daily, leaving the bed when unable to sleep, avoiding naps, using the bed only for sleep and sex • Over-the-counter products: Most of these sleep aids contain antihistamine. They can help you sleep better, but they also may cause severe daytime sleepiness. Other products, including herbal supplements, have little evidence to support their effectiveness. • Prescription sleeping pills: Prescription hypnotics can improve sleep when supervised by a physician. The traditional sleeping pills are benzodiazepine receptor agonists, which are typically prescribed for only short-term use. Newer sleeping pills are , which may pose fewer risks and may be effective for longer-term use. • Unapproved prescription drugs: Drugs from a variety of classes have been used to treat insomnia without FDA approval. Antidepressants such as trazodone are commonly prescribed for insomnia. Others include anticonvulsants, antipsychotics, and nonhypnotic benzodiazepines. Many of these medications involve a significant level of risk.

The American Academy of Sleep Medicine 2510 North Frontage Road Darien, IL 60561 (630) 737-9700 www.aasmnet.org ©AASM 2008

Question 3

Consideration of whether conventional medical therapies are insufficient to treat or alleviate the disease or condition

Contents

Overview – 3

Patients With Treatment-Resistant Insomnia Taking Nightly Prescription Medications for Sleep: A Retrospective Assessment of Diagnostic and Treatment Variables– 4

Diagnosis and Treatment of Chronic Insomnia: A Review– 11

Overview

Insomnia is a sleep disorder that can cause significant impairment to a patient’s daily life. The condition is typically characterized by feelings of fatigue, irritability and memory loss. It is particularly complicated to treat because it can be both a standalone condition and a symptom of other underlying conditions.

Not all patients are able to find relief from insomnia. A study by Barry Krakow, MD and his team (attached below) showed that patients suffering from chronic insomnia in many cases don’t respond to pharmaceuticals. This is of particular note because it demonstrates that a portion on insomnia patients aren’t adequately served by existing treatment options. Confirming this, is a review of available information on insomnia by Ruth Benca, MD. In her review she concludes “Evidence from epidemiologic studies, physician surveys, and clinical studies suggests that numerous patient and physician factors contribute to the fact that the needs of patients with insomnia remain unmet.” Insomnia is a serious condition that can leave a patient feeling helpless. Its vital that physicians be able to diagnose and treat insomnia using any method supported by science. This application, taken on the whole, clearly demonstrates the medical marijuana can be a safe and effective treatment for insomnia.

Patients With Treatment-Resistant Insomnia Taking Nightly Prescription Medications for Sleep: A Retrospective Assessment of Diagnostic and Treatment Variables

Barry Krakow, MD; Victor A. Ulibarri, BS; and Edward A. Romero, BS

Background: Some chronic insomnia patients who take nightly prescription medication achieve less than optimal results. The US Food and Drug Administration (FDA) and the American Academy of Sleep Medicine (AASM) recommend reevaluation of this type of patient to assess for potential psychiatric or medical causes to explain this “failure for insomnia to remit.” Method: A retrospective chart review examined a consecutive series of chronic insomnia patients with persistent insomnia complaints despite current nightly use of prescription medication from May 2005 to February 2008. To assess the role of psychiatric influences on insomnia symptoms, our sample (N = 218) was divided into 2 subgroups: a group with a history of psychiatric complaints (psychiatric insomnia, n = 189) and a control group of no psychiatric complaints (insomnia, n = 29). Results: The average patient reported insomnia for a decade and took prescription medication for sleep for a mean of 4.5 years. Although 100% of the sample used nightly sleep drugs, only 20% believed medication was the best solution for their condition. As evaluated by self-report and polysomnography, these patients exhibited moderately severe insomnia across most measures. Only a few differences were noted between groups. Subjective perception of insomnia severity was worse in the psychiatric insomnia group, which also reported significantly more insomnia-related interference in daily functioning, symptoms of sleep maintenance insomnia, and a trend toward greater daytime fatigue. The mean Apnea-Hypopnea Index score was 19.5 events/hour, yielding an obstructive sleep apnea diagnosis in 75% of patients per conservative AASM nosology (79% in the insomnia group and 74% in the psychiatric insomnia group, P = .22). Conclusions: In this treatment-seeking sample of patients regularly taking sleep medications, residual insomnia was widespread, and patients with psychiatric insomnia may have perceived their condition as more problematic than a control group of insomnia patients without mental health complaints. Both groups exhibited high rates of objectively diagnosed obstructive sleep apnea, a medical condition associated with pervasive sleep fragmentation. These findings support FDA and AASM guidelines to reassess chronic insomnia patients who manifest residual symptoms despite nightly use of prescription medication for sleep.

Prim Care Companion J Clin Psychiatry 2010;12(4):e1–e10 © Copyright 2010 Physicians Postgraduate Press, Inc.

Submitted: August 20, 2009; accepted November 24, 2009. Published online: August 12, 2010 (doi:10.4088/PCC.09m00873bro). Corresponding author: Barry Krakow, MD, Sleep & Human Health Institute, 6739 Academy NE, Ste 380, Albuquerque, NM 87109 ( [email protected]).

hronic insomnia is a common complaint in the general population as well as in various subpopulations such as the elderly,1 psychiatric patients,2,3 and shift C workers4; however, longitudinal data are lacking on their long-term treatment course. Initially, insomnia patients navigate through 4 common pathways: no discernible treatment, over-the-counter sleep aids,5,6 substances or alcohol at bedtime,6,7 and basic sleep hygiene instructions obtained through various media or from primary care providers and educators.8–10 There is often overlap among these 4 pathways. And, in a progression through these approaches, albeit in no fixed order, some insomnia patients broach the issue with a physician or other provider with whom they regularly interact. This type of health care encounter most frequently involves primary care physicians or mental health providers, including psychiatrists, psychologists, and other therapists. In these environments, insomnia patients may receive exposure to evidenced-based treatments for unwanted sleeplessness,8,9,11 for example, prescription medications for sleep and cognitive-behavioral therapies (CBTs). Evidence for CBT as the ideal first-line treatment for insomnia is persuasive and substantial, but the lack of behavioral sleep medicine specialists both at sleep medical centers and in the medical community at large8 has limited the application of this therapeutic option. In contrast, pharmacotherapy for insomnia is well established throughout all fields of medicine. Traditional standards indicate prescribed medication for acute, transient, or situational insomnia, and the prescribing instructions may recommend nightly use for a few weeks or a few times per week for longer intervals.12 However, in clinical settings, it is not unusual for various subgroups of patients, for example psychiatric patients, to rely on the regular, long-term use of prescription medications for sleep. Interestingly, these prescription medications may include standard as well as sedating antidepressants. Indeed, for years trazodone13 was the single most-prescribed medication for sleep, and although there is scant evidence describing the efficacy of antidepressants for insomnia, there can be no doubt that these drugs are often prescribed for the combination of insomnia and depression.14 Recently published American Academy of Sleep Medicine (AASM) guidelines provide support for the use of long-term pharmacologic modalities to treat insomnia.12 However, the guidelines are not a “practice parameter” but rather a “working overview for disease or disorder evaluation and management.”(p490) In fact, when dealing with duration of pharmacologic treatment, it is clearly stated that “the empirical database for long-term treatment remains small.”12(p501) Conversely, government regulatory actions from the US Food and Drug Administration (FDA) established indications for long-term use of the hypnotics eszopiclone, zolpidem controlled release, and ramelteon to treat chronic insomnia.15,16 But, this same agency subsequently posted (March 2007) an important warning about the first 2 medications: “The failure of insomnia to remit after 7 to 10 days of treatment may indicate the presence of a primary psychiatric and/or medical illness that should be evaluated.”17(p3),18(p1) In a similar vein, the AASM practice parameters, which ordinarily do not recommend polysomnography for insomnia patients, state, “… polysomnography is indicated when initial diagnosis (of insomnia) is uncertain, treatment fails (behavioral or pharmacologic), or precipitous arousals occur with violent or injurious behavior.”12(p487) Notwithstanding these paradoxical perspectives, newer agents are now routinely advertised in scientific journals, in popular magazines, on television, and through the Internet as nightly long-term solutions to insomnia. Despite the increase in prescribing patterns for chronic insomnia,19 scant longitudinal data describe how patients respond to these drugs and whether or not FDA and AASM guidelines are invoked subsequent to the “failure of insomnia to remit.” One speculation would comprise a spectrum of patient outcomes, which at one end include responders who continue care with their prescribing physician, whereas those on the opposite end of the spectrum, nonresponders (extreme treatment-resistant cases), might drop out of treatment entirely. In the middle ground, a sizable proportion of partial treatment-resistant insomnia patients manifest mixed results while continuing to use nightly prescription medications. We are interested in this middle group because they commonly present to sleep centers and request further treatment for insomnia despite regularly taking prescription medications. In most cases, these CLINICAL POINTS patients align with AASM and FDA guidelines in that further psychiatric evaluations and polysomnography are indicated to assess residual insomnia complaints. For example, prior studies show Patients with treatment-resistant insomnia taking hypnotic 20–24 higher than expected rates of obstructive sleep apnea (OSA) in chronic insomnia patients. medication require additional medical and psychiatric In the current study, we gathered data on a consecutive series of chronic insomnia patients who evaluations. presented to our sleep medical centers with a history of nightly use of prescription medications for sleep Overnight polysomnography appears to be a useful and the failure of insomnia to remit. From this pool of consecutive patients, a retrospective chart review evaluation tool for patients with treatment-resistant was conducted, coinciding with FDA and AASM guidelines, to assess our sample for psychiatric or insomnia. medical conditions that might be associated with worse insomnia outcomes. For this descriptive study, we developed 2 hypotheses: (1) psychiatric insomnia patients would report or demonstrate worse Sleep-disordered breathing may be common in patients insomnia outcomes than a nonpsychiatric insomnia group and (2) both psychiatric and nonpsychiatric with treatment-resistant insomnia. insomnia groups would show high rates of the medical condition sleep-disordered breathing (SDB) and its resultant sleep fragmentation, which might be associated with residual insomnia symptoms.

METHOD

Per standard protocol at our sleep facilities in Albuquerque and Los Alamos, New Mexico, all patients provide written and verbal consent for their medical information to be used anonymously for research purposes in the context of subsequent chart reviews. This study was a retrospective chart review of existing data on patients treated from May 2005 to February 2008. The study was reviewed and approved by the Los Alamos Medical Center Institutional Review Board, Los Alamos, New Mexico.

Sample and Inclusion Criteria for Chart Review This study contained 4 inclusion criteria: (1) insomnia ranked as the primary reason for seeking treatment despite nightly use of prescription medication for sleep, (2) minimum duration of medication use of 6 months, (3) completion of overnight diagnostic polysomnography or split-night protocol with at least 2 hours of diagnostic sleep data, and (4) at least 18 years of age. Figure 1 (flowchart) shows nearly half the clinic population of 2,236 patients ranked insomnia as their primary problem, but only 301 patients reported both residual insomnia and nightly sleep medication use. Each individual patient’s medications were reviewed to determine whether or not their reported nightly agent was in fact being used to treat sleep problems. For example, 3 patients were excluded because the only nighttime medication was lithium, a drug not typically used for insomnia. Additional exclusions were duration of use less than 6 months, no polysomnography testing, or age less than 18 years. And, as anticipated from our introduction, no patients presented to our sleep medical facility with a self-reported optimal response to nightly use of prescription medication for sleep.

Psychiatric Assessment No formal psychiatric interviewing or instruments were used to evaluate historical complaints of psychiatric conditions; therefore, the information used in this retrospective study relies entirely on subjective reports from the patients’ intake questionnaires. However, the intake questionnaire does not ask patients to speculate about their mental health; rather, it specifically asks patients to check those conditions that they believe they suffer from currently or have suffered from in the past. The checklist comprises 4 columns: a list of 8 psychiatric conditions and 2 additional lines for “none” or “other,” a list of current medications taken for a condition checked, a list of past medications taken for the condition, and any other therapies used to treat the condition. Two additional questions are asked about a history of traumatic exposure and/or claustrophobia. For trauma, patients are asked if they have suffered a threat to their life, a serious injury, or a serious assault (physical, sexual, etc), which elicited a feeling of fear, helplessness, and/or horror. Although these data do not corroborate formal psychiatric diagnoses, 87% of the patients in our sample reported a history of at least 1 of the following: depression, anxiety disorder, posttraumatic stress disorder CLICK FIGURE TO ENLARGE (PTSD), panic disorder, schizophrenia, bipolar disorder, obsessive-compulsive disorder (OCD), traumatic exposure, or claustrophobia. To be clear, however, these data are only describing the association of a putative psychiatric condition with the presence of insomnia; the data do not indicate whether or not the psychiatric condition is a cause or contributor to the patient’s insomnia. Therefore, in this article, the term psychiatric insomnia is only used to distinguish this group of patients from those with no self-reported psychiatric condition(s).

Sleep Intake, Measurements, and Procedures All patients seen at both sleep facilities completed an online intake set of questionnaires, including an extensive sleep medicine history25 on self-reported subjective sleep measures, sleep history, and medication use for sleep. The sleep medicine history also contained 3 questions about patient perceptions on chronic use of sedating medications, and these were scored on a disagree/unsure/agree 3-point scale. These questions dealt with whether medication was perceived as the best solution for their sleep problems, whether they had been seeking help for their sleep problems for an extended period of time, and the level of frustration with past efforts in getting help for their sleep problem(s). Answers to these questions were not considered for inclusion or exclusion in the current study. The Insomnia Severity Index (ISI)26 was also completed by all patients. The ISI is the sum of 7 individual questions dealing with the following: (1) difficulty falling asleep, (2) difficulty staying asleep, (3) waking up too early, (4) satisfaction with current sleep pattern, (5) extent of interference sleep problem has on daily functioning, (6) impairment of quality of life noticeable to others, and (7) level of distress surrounding current sleep problem. Each question is scored on a scale of 0 to 4 based on increasing severity of the symptom, and the total score ranges from 0 to 28 with scores greater than 20 equal to moderately severe to severe insomnia. Two visual analog scales (VAS) were used to measure daytime sleepiness and tiredness,27,28 scored on a 0 to 10 metric with scores > 6 indicative of severe symptoms. Intake questionnaires are reviewed by the sleep specialist (B.K.), and patients are scheduled for polysomnography when criteria are met for apparent risks of physiologic disorders such as SDB. In our clinical experience, we follow FDA and AASM guidelines and recommend diagnostic polysomnography testing in the large majority of these patients17–18 for failing to achieve a satisfactory or optimal response to pharmacologic treatment of insomnia. However, as noted inFigures 1, 40 patients refused or did not undergo polysomnography testing and were therefore excluded from the study.

Polysomnography Protocol Overnight sleep studies were performed using standard polysomnography on all Maimonides Sleep Arts & Sciences and Los Alamos Medical Center Sleep Laboratory patients. Technicians prepared patients using the international 10–20 system of electrode placement. The recording had a 14-channel montage: left outer canthus-A2 , right outer canthus-A132411221 , C -A , C -A , O -A , O -A , chin, electrocardiogram, left leg-right leg, snore, nasal pressure transducer via nasal canula, chest effort, abdominal effort, pulse oximetry, and position. Polysomnography sleep staging was scored manually by registered technologists using Rechtschaffen and Kales’29 scoring criteria. Three types of events were scored. An apnea was a > 70% decrease in airflow for at least 10 seconds. Hypopnea was a 30%–70% decrease in airflow coupled with either a 4% oxygen desaturation or an arousal. Flow limitation was a decrease in airflow of ≤ 30% in the form of classic flattening or notching of the airflow limb for at least 2 consecutive breaths, lasting > 10 seconds, ending in an arousal. Minimum oxygen saturation was recorded by pulse oximetry. All patients were tested using their regular bedtime regimen of medications. The Apnea-Hypopnea Index (AHI) was calculated for each patient and per AASM nosology.30 An AHI score ≥ 5 was used as a diagnostic cut point for the diagnosis of SDB. A total of 18 patients completed split nights, and their data were extrapolated as if they completed a full-night study by using their objective sleep efficiency from the diagnostic portion of the polysomnography testing to estimate relevant objective sleep indices.

Data Analysis To compare outcomes among those patients with and without self-reported psychiatric conditions, 2 groups were created: insomnia group (no psychiatric complaints, n = 29) and psychiatric insomnia group (self-report of psychiatric complaints, n = 189). One post hoc analysis compared patients with OSA (n = 163) and no OSA (n = 55). Analysis of variance and χ2 analyses were performed for continuous and dichotomous variables, respectively. Cohend effect sizes, the standardized mean difference, were calculated from the difference in 2 means, divided by the pooled standard deviation, and used for select comparisons of statistically significant findings only. Statistical significance was .05. Power analysis (α = .5, β = .80) revealed that the 2 unequal samples for the insomnia and psychiatric insomnia groups were slightly underpowered to detect medium-sized effects.

RESULTS

Sample Characteristics Of the 2 contrasted groups, the psychiatric insomnia group comprised 189 patients who reported the following prevalence of psychiatric conditions: depression (56.4%), traumatic exposure (50.2%), anxiety (42.7%), claustrophobia (41.6%), panic (25.7%), PTSD (17.95%), bipolar illness (9.2%), OCD (2.8%), disorder (3.2%), and schizophrenia (0.5%). Table 1 shows that the full sample of unresolved insomnia patients was largely middle-aged and slightly overweight. Distinctions between groups revealed predominantly women in the psychiatric insomnia group and predominantly men in the insomnia group (P = .004). Insomnia severity, based on ISI score, was in the moderate to severe range (mean ± SD = 19.16 ± 5.63) with the psychiatric insomnia group having a significantly higher ISI (19.52 ± 5.37) than the insomnia group (16.68 ± 6.75) (P = .01, d = 0.47). The mean ± SD duration of insomnia was 12.2 ± 13.32 years. Sedating medication use ranged from 6 months to 30 years with a mean of 54.68 months (SD = 69.84) and a median of 25.0 months.

CLICK FIGURE TO ENLARGE

Prescription Medications for Sleep All patients were using a prescription medication designated by their prescribing physician for the primary purpose of treating a sleep disturbance. Of our total sample, 67% were prescribed sleep medications by their primary care physician, 24% by a psychiatrist, 8% by physicians of various specialties, and 1% by a sleep doctor. However, due to patient inconsistency, we did not collect or report specific dosages used by the patients in this sample. We identified 3 main categories of drugs prescribed (Table 2): benzodiazepines (n = 69), nonbenzodiazepines (n = 75), and mood stabilizers/antidepressants (n = 74). No patients were taking the melatonin agonist ramelteon. In the psychiatric insomnia group, roughly one-half of the patients were also taking other psychotropic medications that influence sleep,31,32 although these drugs were not the primary medication prescribed for sleep. No systematic differences were evident between these 3 prescribed groups of medications for any of the variables of interest in this study; therefore, no additional analyses were conducted on the basis of medications.

Patient Perceptions CLICK FIGURE TO ENLARGE On the basis of 3 questions about patient perceptions, 74% of the total sample reported that they had been seeking help for their sleep problems for a long time, which is consistent with a mean duration of insomnia of 12 years. Nearly half the sample (47%) reported frustration with previous physicians in their attempts to solve their sleep problems, whereas only 26% reported no frustration and 27% were unsure how to respond. At this point in their care, only 20% of patients believed that the correct medication would solve their sleep problems, whereas 55% were undecided on this question, and 24% believed medications would not solve their problems. There were no statistical differences on these perceptions between the insomnia and psychiatric insomnia groups. Two individual items on the ISI relating to patient perceptions showed differences between the 2 groups. First, those patients in the psychiatric insomnia group reported more difficulty staying asleep (P = .05, d = 0.36). Second, the psychiatric insomnia group also perceived that their sleep problems interfered with their daily functioning to a greater degree than the insomnia group (P = .04, d = 0.38), and these 2 items account for most of the differences between the 2 groups on the total ISI score of self-reported insomnia severity. Regarding impairment, there were no significant differences among groups for the VAS sleepiness scale (Table 1); however, the VAS for tiredness showed that the psychiatric insomnia group trended toward greater severity than the insomnia group (P = .08, d = 0.32).

Self-Reported and Objective Sleep Indices Subjective sleep indices including mean ± SD sleep-onset latency (69.11 ± 65.67 minutes), total sleep time (6.01 ± 1.93 hours), wake time after sleep onset (100.43 ± 89.72 minutes), and calculated sleep efficiency (73.24 ± 19.98%) consistently showed moderate to severe insomnia. Patients also reported other signs of sleep fragmentation including difficulty maintaining sleep, increased awakenings, and difficulties returning to sleep once awakened. There were no statistical differences between the 2 groups. For objective sleep indices, the polysomnography testing consisted of 1 night in the sleep laboratory, and the patients used their standard nightly regimen of medications including their primary sleep medication. As expected, consistent signs of moderate to severe insomnia were documented for total sleep time, wake time after sleep onset, and sleep efficiency without significant differences between the 2 groups (Table 3). And, as often seen with first-night effects in sleep laboratory testing of insomnia patients,33 mean ± SD sleep-onset latency was within normal limits in both groups (insomnia group: 11.90 ± 9.51, psychiatric insomnia group: 20.69 ± 35.11, P = .18). When looking at individual sleep stage ratios (Table 3), the psychiatric insomnia group showed a significantly higher ratio (66.41 ± 13.38) of stage 2 non–rapid eye movement (NREM) sleep than the insomnia group (59.20 ± 17.87, P = .01) with a medium effect size (d = 0.46). The ratio of stage 1 NREM sleep trended lower in the psychiatric insomnia group (14.25 ± 11.26) versus the insomnia group (18.68 ± 17.29, P = .07, d = 0.30), whereas the ratio of stage 4 NREM sleep in the psychiatric insomnia group trended lower (1.40 ± 4.32) than the insomnia group (3.11 ± 8.19, P = .09, d = 0.26).

CLICK FIGURE TO ENLARGE

Breathing Event Indices and Sleep-Disordered Breathing Diagnoses The mean ± SD AHI score was 19.48 ± 23.62 events/hour with no significant differences between the 2 groups (Table 4). Given a significant difference in men versus women between groups (more men in the insomnia group) and the consequent greater severity of OSA (AHI) found in men (men: 29.83 ± 29.76, women: 12.34 ± 14.52, P =

.001), further analysis showed that the finding of more severe OSA in the insomnia group was primarily due to differences in gender ratios between groups (F2,235 = 8.210, P = .0001).

CLICK FIGURE TO ENLARGE

According to AASM nosology,30 we found that 163 (75.0%) patients met criteria for OSA based solely on AHI score (≥ 5) (79% of the insomnia group and 74% of the psychiatric insomnia group). For exploratory purposes only and not reported in our tables, we scored flow limitation events 2 ways: (1) according to AASM guidelines,30 yielding a mean ± SD Respiratory Disturbance Index (RDI) of 37.54 ± 24.83 events/hour, and (2) with so-called “subcortical arousals” using Rapoport’s criteria,31 which would have raised the mean ± SD RDI to 46.53 ± 25.24 events/hour, thereby potentially yielding an even higher proportion of diagnosed cases of OSA. There was no significant difference of mean RDI values between the 2 groups.

Sleep Fragmentation We assessed sleep fragmentation in the form of awakenings (> 15.0 seconds), arousals (3.0–15.0 seconds), and microarousals (1.5–3.0 seconds) caused by sleep breathing events that might be associated with insomnia outcomes (Table 4). For both groups, SDB events produced a mean of 1 awakening/hour due to apneas (including central apneas) or hypopneas with an additional 1.7 awakenings/hour for flow limitation events (Table 5). The Respiratory Arousal Index showed a mean of 14.21 ± 20.04 arousals/hour for apneas and hypopneas with an increase to 28.60 ± 16.11 arousals/hour when flow limitation events were included. There was a trend toward a higher total of SDB-related arousals in the insomnia group (34.61 ± 26.74) when compared to the psychiatric insomnia group (28.29 ± 15.02, P = .07, d = 0.29). And finally, OSA caused 2.68 ± 10.32 microarousals/hour with an increase to 6.28 microarousals/hour when flow limitation events were included. However, none of these awakenings, arousals, and microarousals individually or in combination correlated with ISI severity. CLICK FIGURE TO ENLARGE

OSA Versus Non-OSA Patients A final analysis was performed comparing those patients meeting AASM nosology for the diagnosis of OSA (AHI ≥ 5) and those who did not (AHI < 5). Significant or trending differences are found in Table 5. In sum, OSA patients were older and more likely to be men and overweight. Patients with OSA were more likely to report classic breathing symptoms including snoring, history of moving from the bed because of patient’s snoring, witnessed apneas, and choking, gasping, or struggling for breath during sleep. Subjective sleep indices showed the following: OSA patients trended toward shorter sleep-onset latency, longer total sleep time, and more consolidated sleep efficiency. Interestingly, insomnia severity based on mean ± SD total ISI score was worse in the non-OSA group (19.52 ± 5.37) than in the OSA group (16.68 ± 6.75, P = .004, d = 0.47). Objective measures show expected results of worse values in the OSA group compared to the non-OSA group Table( 5). Last, we compared all of our subjective sleep data for the 40 individuals who were excluded for not undergoing polysomnography testing with those 218 patients in our study, and we found no significant differences on any variable between the 2 groups.

DISCUSSION

In this retrospective chart review, there were 3 primary findings in this consecutive series of treatment-resistant chronic insomnia patients who averaged reports of insomnia for more than a decade and regular use of prescription medication for sleep for 4 years. First, the findings of moderate to severe residual insomnia clearly indicated a failure to respond adequately to sleep medications. Second, patients reporting psychiatric problems were far more common than insomnia patients without such complaints, but there were only a few differences between these groupings in contrast to what we had predicted. Third, objective polysomnography diagnosed obstructive sleep apnea in three-fourths of this select sample along with the innate sleep fragmentation that invariably accompanies this disorder of sleep respiration. However, despite the potential importance of these findings, no systematic links unequivocally connected psychiatric influences or SDB to the etiology or severity of the residual insomnia reported by this sample. Regarding psychiatric influences, the only distinctions were that the psychiatric insomnia group believed their sleep maintenance insomnia and daytime impairment were worse than the insomnia group, and the psychiatric insomnia group showed a trend toward worse daytime fatigue. Similarly, there were only a few differences with respect to objective findings on sleep stage comparisons; but these findings might be related to greater use of psychotropic medications in the psychiatric insomnia group.32,34 The most relevant objective finding was that the psychiatric insomnia group (comprised of more women) showed a greater severity of flow limitation or upper airway resistance compared to the insomnia group (comprised of more men) that showed greater apneas and hypopneas. Regarding medical influences on chronic insomnia, the current study revealed that many of these patients suffered from SDB, the most common medical cause of sleep fragmentation,20 and among the OSA group, 75% reported snoring and one-third of these patients reported various breathing symptoms suggestive of OSA. Yet, it remains unclear to what extent these breathing symptoms had been discussed or evaluated prior to seeking care at our sleep medical center. Our lack of a longitudinal design prevents us from knowing whether use of sleeping pills worsened OSA through respiratory depressant effects35 or whether these medications ameliorated some of the sleep-fragmenting effects of OSA by raising arousal thresholds.36 Also, as the study was conducted at 5,200 ft above sea level, altitude may be a contributing factor to increased OSA and central sleep apnea events. We could speculate that therapeutic application of positive airway pressure therapy, the gold standard treatment for SDB, would influence insomnia outcomes or alter medication use or dosage, but there are only a few weakly designed studies20,37,38 that report improved insomnia outcomes subsequent to the treatment of OSA. The current findings are consistent with prior research showing higher than anticipated rates of comorbid insomnia and OSA (“complex insomnia”)39 in a sample of patients who would ordinarily not be expected to suffer from a sleep breathing disorder. Of some potential clinical interest, insomnia severity was less in the OSA group compared to those without OSA, and objective measures on polysomnography testing tended to corroborate this finding, which in our view, further highlights the elusive nature of the pathophysiologic relationship between these 2 common sleep disorders. The many limitations of this design mandate a cautious interpretation of our findings. Most importantly, this sample may prove to be highly select in that they sought treatment because medications were not working for them, and a lack of response to treatment is a well-known motivator to seek additional help or second opinions when the medical condition is vexing enough to move the patient to action.40 However, due to the design of this study, we cannot state that psychiatric influences or SDB were the causes of the patients’ residual insomnia complaints; we can only state that these influences were present to some degree but are of unknown clinical significance. Moreover, for psychiatric influences, we only used self-report checklists and did not conduct psychiatric interviews or use psychiatric instruments to make formal diagnoses. As such, our findings on psychiatric influences only suggest information about general distress not about specific diagnostic categories. And, there was no longitudinal data on the interaction between the patients’ psychiatric and insomnia treatments, including psychotherapies or medications, so the snapshot design of the protocol limits our understanding of what transpired with these treatment-resistant insomnia patients prior to their arrival at the sleep center. Along the same lines, 1 night of polysomnography also reflects a snapshot of sleep stages and breathing events. Again, a longitudinal design with more frequent testing and research-oriented polysomnography protocols as well as the utilization of sleep diaries would likely yield more useful clinical data. Regarding our 2 groups, our sample of nonpsychiatric insomnia patients was notably smaller and contained a greater proportion of men with only a few differences between the 2 groups, all of which might have changed with larger more balanced samples. Last, we gathered no reliable data on specific drug dosages for insomnia used by these patients, which clearly could have influenced outcomes given that many patients do not take their medications consistently as prescribed.41,42 Moreover, the psychiatric insomnia group consistently and significantly was taking multiple psychotropic drugs compared to the insomnia group. Overall, our findings may not be generalizable to other chronic insomnia patients taking various sleep medications on a long-term basis, particularly in light of the successes that have been described in the research literature for the long-term use of newer medications.

CONCLUSIONS

To our knowledge, this is the first study to examine pertinent outcomes in a sample of chronic insomnia patients maintaining nightly prescription medication for sleep despite clear-cut signs of treatment-resistant insomnia. The scientific literature does not contain much information about such patients perhaps due to a lack of interest or reluctance to look at treatment outcomes in this difficult patient population.43 Nonetheless, our study demonstrated the value in both the AASM and FDA guidelines,12,17,18 because our findings suggest psychiatric and medical factors may be associated with patients who are not responding optimally to therapy. In particular, the sleep fragmentation effects of both psychiatric and medical conditions, such as those described in this study, may prove to be important influences in these cases. Although it is tempting to imagine that additional assessment and treatments for either sleep breathing or psychiatric conditions would lead to greater improvement in insomnia severity among this select sample of patients, only randomized controlled trials with evidence-based therapies can address such speculations.

Drug names: alprazolam (Niravam, Xanax, and others), bupropion (Aplenzin, Wellbutrin, and others), clonazepam (Klonopin and others), diazepam (Diastat, Valium, and others), escitalopram (Lexapro and others), eszopiclone (Lunesta), (Prozac and others), flurazepam (Dalmane and others), haloperidol (Haldol and others), lorazepam (Ativan and others), (Zyprexa), paroxetine (Pexeva, Paxil, and others), ramelteon (Rozerem), sertraline (Zoloft and others), temazepam (Restoril and others), venlafaxine (Effexor and others), zaleplon (Sonata and others), zolpidem (Ambien, Zolpimist, and others). Author affiliations: Sleep and Human Health Institute, Maimonides Sleep Arts & Sciences, Ltd (all authors); and Los Alamos Medical Center Sleep Laboratory (Dr Krakow), Albuquerque, New Mexico. Potential conflicts of interest: Dr Krakow owns and operates 3 Web sites that provide education and offer products and services for patients with sleep disorders ( http://www.nightmaretreatment.com, http://www.sleeptreatment.com, and http:www.sleepdynamictherapy.com); markets and sells 3 books for patients with sleep disorders ( Insomnia Cures, Turning Nightmares Into Dreams, and Sound Sleep, Sound Mind); owns and operates Maimonides Sleep Arts & Sciences, Ltd; and serves as president of a nonprofit sleep research center, the Sleep & Human Health Institute, which occasionally provides consultation services. Messrs Ulibarri and Romero report no financial or other affiliations relevant to the subject of this article. Funding/support: None reported.

REFERENCES

1. Ring D. Management of chronic insomnia in the elderly. Clin Excell Nurse Pract. 2001;5(1):13–16. PubMed 2. Szelenberger W, Soldatos C. Sleep disorders in psychiatric practice. World Psychiatry. 2005;4(3):186–190. PubMed 3. Maher MJ, Rego SA, Asnis GM. Sleep disturbances in patients with post-traumatic stress disorder: epidemiology, impact and approaches to management.CNS Drugs. 2006;20(7):567–590. PubMed doi:10.2165/00023210-200620070-00003 4. Schwartz JR, Roth T. Shift work sleep disorder: burden of illness and approaches to management.Drugs . 2006;66(18):2357–2370. PubMed doi:10.2165/00003495- 200666180-00007 5. Kupfer DJ, Reynolds CF III. Management of insomnia.N Engl J Med. 1997;336(5):341–346. PubMed doi:10.1056/NEJM199701303360506 6. Morin CM, Beaulieu-Bonneau S, LeBlanc M, et al. Self-help treatment for insomnia: a randomized controlled trial.Sleep . 2005;28(10):1319–1327. PubMed 7. Roehrs T, Papineau K, Rosenthal L, et al. Ethanol as a hypnotic in insomniacs: self administration and effects on sleep and mood.Neuropsychopharmacology . 1999;20(3):279–286. PubMed doi:10.1016/S0893-133X(98)00068-2 8. Murtagh DR, Greenwood KM. Identifying effective psychological treatments for insomnia: a meta-analysis. J Consult Clin Psychol. 1995;63(1):79–89. PubMed doi:10.1037/0022-006X.63.1.79 9. Holbrook AM, Crowther R, Lotter A, et al. The diagnosis and management of insomnia in clinical practice: a practical evidence-based approach.CMAJ . 2000;162(2):216–220. PubMed 10. Morin AK. Strategies for treating chronic insomnia. Am J Manag Care. 2006;12(suppl 8):S230–S245. PubMed 11. Smith MT, Perlis ML, Park A, et al. Comparative meta-analysis of pharmacotherapy and behavior therapy for persistent insomnia.Am J Psychiatry. 2002;159(1):5–11. PubMed doi:10.1176/appi.ajp.159.1.5 12. Schutte-Rodin S, Broch L, Buysse D, et al. Clinical guideline for the evaluation and management of chronic insomnia in adults.J Clin Sleep Med. 2008;4(5):487–504. PubMed 13. Brooks D, Prothero W, Bouras N, et al. Trazodone: a comparison of single night-time and divided daily dosage regimens.Psychopharmacology (Berl). 1984;84(1):1–4. PubMed doi:10.1007/BF00432013 14. Becker PM. Treatment of sleep dysfunction and psychiatric disorders. Curr Treat Options Neurol. 2006;8(5):367–375. PubMed doi:10.1007/s11940-006-0026-6 15. US Food and Drug Administration. Ambien CR approval letter. 2005 http://www.fda.gov/cder/foi/appletter/2005/021774ltr.pdf. Accessed June 2008. 16. US Food and Drug Administration. Lunesta approval letter. 2008. http://www.fda.gov/cder/foi/appletter/2004/021476ltr.pdf. Accessed June 2008. 17. US Food and Drug Administration. Ambien. Highlights of prescribing information. http://www.accessdata.fda.gov/drugsatfda_docs/label/2007/019908s022lbl.pdf. Accessed July 2010. 18. US Food and Drug Administration. Lunesta (eszopiclone) tablets.http://www.fda.gov/Safety/MedWatch/SafetyInformation/Safety- RelatedDrugLabelingChanges/ucm106688.htm. Accessed July 2010. 19. Botteman MF, Ozminkowski RJ, Wang S, et al. Cost effectiveness of long-term treatment with eszopiclone for primary insomnia in adults: a decision analytical model. CNS Drugs. 2007;21(4):319–334. PubMed doi:10.2165/00023210-200721040-00005 20. Guilleminault C, Palombini L, Poyares D, et al. Chronic insomnia, premenopausal women and sleep disordered breathing, part 2: comparison of nondrug treatment trials in normal breathing and UARS post menopausal women complaining of chronic insomnia. J Psychosom Res. 2002;53(1):617–623. PubMed doi:10.1016/S0022- 3999(02)00463-4 21. Krakow B, Melendrez D, Warner TD, et al. Signs and symptoms of sleep-disordered breathing in trauma survivors: a matched comparison with classic sleep apnea patients. J Nerv Ment Dis. 2006;194(6):433–439. PubMed doi:10.1097/01.nmd.0000221286.65021.e0 22. Krakow BJ, Melendrez DC, Johnston LG, et al. Sleep dynamic therapy for Cerro Grande fire evacuees with posttraumatic stress symptoms: a preliminary report.J Clin Psychiatry. 2002;63(8):673–684. PubMed 23. Krakow B, Melendrez D, Pedersen B, et al. Complex insomnia: insomnia and sleep-disordered breathing in a consecutive series of crime victims with nightmares and PTSD. Biol Psychiatry. 2001;49(11):948–953. PubMed doi:10.1016/S0006-3223(00)01087-8 24. Sivertsen B, Omvik S, Pallesen S, et al. Sleep and sleep disorders in chronic users of and drug-free insomniacs.J Clin Sleep Med. 2009;5(4):349–354. PubMed 25. Krakow B, Melendrez D, Ferreira E, et al. Prevalence of insomnia symptoms in patients with sleep-disordered breathing.Chest . 2001;120(6):1923–1929. PubMed doi:10.1378/chest.120.6.1923 26. Bastien CH, Vallières A, Morin CM. Validation of the Insomnia Severity Index as an outcome measure for insomnia research.Sleep Med. 2001;2(4):297–307. PubMed doi:10.1016/S1389-9457(00)00065-4 27. Lee KA, Hicks G, Nino-Murcia G. Validity and reliability of a scale to assess fatigue.Psychiatry Res. 1991;36(3):291–298. PubMed doi:10.1016/0165-1781(91)90027-M 28. Zallek SN, Redenius R, Fisk H, et al. A single question as a sleepiness screening tool.J Clin Sleep Med. 2008;4(2):143–148. PubMed 29. Rechtschaffen A, Kales A. A Manual of Standardized Terminology, Techniques and Scoring System for Sleep Stages of Human Subjects. Los Angeles, CA: Brain Information Service/Brain Research Institute; 1968. 30. American Academy of Sleep Medicine. International Classification of Sleep Disorders: Diagnostic and Coding Manual. 2nd ed. Westchester, IL: American Academy of Sleep Medicine; 2005. 31. Hosselet J, Ayappa I, Norman RG, et al. Classification of sleep-disordered breathing. Am J Respir Crit Care Med. 2001;163(2):398–405. PubMed 32. Oberndorfer S, Saletu-Zyhlarz G, Saletu B. Effects of selective serotonin reuptake inhibitors on objective and subjective sleep quality.Neuropsychobiology . 2000;42(2):69–81. PubMed doi:10.1159/000026676 33. Riedel BW, Winfield CF, Lichstein KL. First night effect and reverse first night effect in older adults with primary insomnia: does anxiety play a role?Sleep Med. 2001;2(2):125–133. PubMed doi:10.1016/S1389-9457(00)00054-X 34. Sharpley AL, Cowen PJ. Effect of pharmacologic treatments on the sleep of depressed patients.Biol Psychiatry. 1995;37(2):85–98. PubMed doi:10.1016/0006- 3223(94)00135-P 35. Hedemark LL, Kronenberg RS. Flurazepam attenuates the arousal response to CO2 during sleep in normal subjects.Am Rev Respir Dis. 1983;128(6):980–983. PubMed 36. Lu B, Budhiraja R, Parthasarathy S. Sedating medications and undiagnosed obstructive sleep apnea: physician determinants and patient consequences.J Clin Sleep Med. 2005;1(4):367–371. PubMed 37. Krakow B, Melendrez D, Sisley B, et al. Nasal dilator strip therapy for chronic sleep maintenance insomnia: a case series.Sleep Breath. 2004;8(3):133–140. PubMed doi:10.1055/s-2004-834483 38. Krakow B, Melendrez D, Lee SA, et al. Refractory insomnia and sleep-disordered breathing: a pilot study.Sleep Breath. 2004;8(1):15–29. PubMed doi:10.1007/s11325- 004-0015-5 39. Benetó A, Gomez-Siurana E, Rubio-Sanchez P. Comorbidity between sleep apnea and insomnia.Sleep Med Rev. 2009;13(4):287–293. PubMed doi:10.1016/j.smrv.2008.09.006 40. Meyer D, Leventhal H, Gutmann M. Common-sense models of illness: the example of hypertension.Health Psychol. 1985;4(2):115–135. PubMed doi:10.1037/0278- 6133.4.2.115 41. Cooper C, Bebbington P, King M, et al. Why people do not take their psychotropic drugs as prescribed: results of the 2000 National Psychiatric Morbidity Survey.Acta Psychiatr Scand. 2007;116(1):47–53. PubMed doi:10.1111/j.1600-0447.2006.00974.x 42. Mitchell AJ. Adherence behaviour with psychotropic medication is a form of self-medication.Med Hypotheses. 2007;68(1):12–21. PubMed doi:10.1016/j.mehy.2006.07.005 43. Kramer M. Hypnotic medication in the treatment of chronic insomnia: non nocere! Doesn’t anyone care?Sleep Med Rev. 2000;4(6):529–541. PubMed doi:10.1053/smrv.2000.0122 Diagnosis and Treatment of Chronic Insomnia: A Review Ruth M. Benca, M.D., Ph.D.

Objective: Insomnia has high prevalence rates and is associated with them with current information about significant personal and socioeconomic burden, yet it remains largely pharmacologic and nonpharmacolog- underrecognized and inadequately treated. Methods: A PubMed search ic treatments for insomnia. for English-language articles covering randomized controlled trials published between 1970 and 2004 was conducted. Search terms used Methods were “insomnia,” “behavioral therapy,” and the generic names of A PubMed search for English-lan- agents commonly used to treat insomnia (the Food and Drug Adminis- guage articles published between tration–approved benzodiazepines and nonbenzodiazepines, tra- 1970 and 2004 that reported random- zodone, and over-the-counter agents). Results: Evidence from epi- ized controlled trials or active com- demiologic studies, physician surveys, and clinical studies suggests that parator trials was conducted. The numerous patient and physician factors contribute to the fact that the search terms “insomnia” and “behav- needs of patients with insomnia remain unmet, including low reporting ioral therapy” were used to search for of insomnia by patients, limited physician training, and office-based articles on nonpharmacologic treat- time constraints, as well as misconceptions about the seriousness of in- ments. For articles on pharmacologic somnia, the advantages of treatment, and the risks associated with treatments the search terms used hypnotic use. Nonpharmacologic therapies produce long-lasting and were “insomnia” and “flurazepam,” reliable changes among people with chronic insomnia and have mini- “,” “estazolam,” “temaze- mal side effects. Pharmacologic therapies have proven effective with pam,” “triazolam” (Food and Drug improving wake time after sleep onset and sleep maintenance and re- Administration [FDA]–approved ben- ducing the number of nighttime awakenings. However, pharmacologic zodiazepines), “zolpidem,” and “zale- therapy has a greater chance of producing side effects. No conclusive plon” (FDA-approved nonbenzodi- evidence exists to favor either pharmacologic therapy or behavioral azepines), as well as “trazodone” therapy. Conclusions: Insomnia is particularly challenging for clini- (commonly used for treating insom- cians because of the lack of guidelines and the small number of studies nia, although not FDA-approved for conducted in patient populations with behavioral and pharmacologic this purpose) and “diphenhydra- therapies. Current treatment options do not address the needs of diffi- mine” (also commonly used for treat- cult-to-treat patients with chronic insomnia, such as the elderly, and ing insomnia). those with comorbid medical and psychiatric conditions. More re- For nonpharmacologic treat- search is necessary to determine the long-term effects of insomnia ments, a total of 11 papers were treatments. (Psychiatric Services 56:332–343, 2005) identified that discussed cognitive- behavioral, sleep-restriction, and sleep hygiene therapies. In total, 41 t is estimated that 10 to 15 per- nized, underdiagnosed, and under- studies were identified for approved cent of the adult population suf- treated (1). benzodiazepines. Of these, most Ifers from chronic insomnia; an Mental health care professionals were studies in which an active com- additional 25 to 35 percent has tran- frequently treat patients who have in- parator was used and included pa- sient or occasional insomnia (1). somnia that co-occurs with a psychi- tients with primary and secondary Fifty-seven percent of noninstitution- atric disorder. A literature review was insomnia. Eight nonbenzodiazepine alized elderly persons experience undertaken to help clinicians better studies, all of which included a chronic insomnia (2). Despite these understand the nature of insomnia placebo comparison, were identi- high prevalence rates, evidence sug- and the reasons for its underdiagnosis fied. Eighteen studies evaluated tra- gests that insomnia is underrecog- and undertreatment and to provide zodone’s efficacy on sleep endpoints, and in all except two (3,4) of these the sample was limited to patients Dr. Benca is a professor in the department of psychiatry at the University of Wisconsin, with depression- or antidepressant- 6001 Research Park Boulevard, Madison, Wisconsin 53719 (e-mail, rbenca@med. induced insomnia. Six studies evalu- wisc.edu). ated diphenhydramine.

332 PSYCHIATRIC SERVICES ♦ http://ps.psychiatryonline.org ♦ March 2005 Vol. 56 No. 3 Results Barriers to recognition, diagnosis, and treatment Barriers to recognition and treatment of insomnia Several factors hinder the appropri- Inadequate physician training in insomnia ate recognition of insomnia and its Time-constrained physician office visits adequate and appropriate manage- Lack of discussion about sleep problems during patient consultations ment. They include lack of physician Belief among patients and physicians that sleep complaints are not education about insomnia, time-con- important strained patient visits, beliefs among Perception by physicians that treatments for insomnia are ineffective or associated with risks patients and physicians that sleep Lack of evidence that treating insomnia improves outcomes of comorbid complaints are not important, the be- conditions lief that treatments are not effective or cause more problems, and the lack of research evidence that treating in- somnia improves outcomes of comor- bid conditions (listed in the box on avoid having to deal with issues that prescribing hypnotic medications, this page). could take up more than the normal and the perceived risks associated Physician training. Most clini- allotted time for a patient. with use of these agents. Historically, cians are not well trained with respect Reporting sleep problems or studies that have evaluated medica- to sleep and sleep disorders. A survey eliciting a sleep history. Deficits in tions in randomized controlled trials conducted from 1990 to 1991 indicat- knowledge about sleep medicine and of insomnia lasting more than six ed that 37 percent of medical schools time constraints both contribute to the weeks are rare; a meta-analysis of were still failing to include structured reticence among physicians to tackle studies using benzodiazepines and sleep medicine sessions in their cur- issues related to sleep. In the World zolpidem published in 1997 demon- ricula. On average, less than two hours Health Organization’s international strated that median duration of trials of total teaching time was allocated to collaborative study on general health was seven days (11). Until recently, sleep and sleep disorders, and only 11 care, attendees in 15 primary care set- little was known, therefore, about ef- percent of medical students partici- tings in several countries found that ficacy and safety beyond this duration pated in the clinical evaluation of pa- physicians detected insomnia in less of treatment, despite the fact that tients with sleep disorders (5). than 50 percent of patients who had many patients may require longer- Although research in this area is insomnia symptoms (8). A survey by term treatment. FDA labeling states, limited, studies have demonstrated Papp and colleagues (6) found that “Hypnotics should generally be limit- that the lack of training in sleep disor- time spent counseling patients on the ed to seven to ten days of use, and ders is reflected in knowledge deficits benefits of sleep was less than that reevaluation of the patient is recom- about sleep medicine among primary spent discussing diet or exercise. mended if they are to be taken for care physicians (6). In a study that as- Neglect of sleep difficulties ap- more than two to three weeks. [The sessed the sleep knowledge of 580 pears to partially be the result of re- drug] should not be prescribed in primary care physicians, none rated luctance on the parts of both patients quantities exceeding a one-month their knowledge of sleep disorders as and physicians to discuss sleep. Ford supply” (12). These regulations re- excellent; 10 percent rated their and Kamerow (10) found that of strict prescription of hypnotics to knowledge as good, 60 percent as fair, those in their cohort who described short-term treatments and make the and 30 percent as poor (6). difficulty sleeping only 9 percent re- chronic prescription of sedative-hyp- Office visits. The advent of a man- ported the problem to a physician. notics difficult. aged-care-based clinical environment Shochat and colleagues (9) found that FDA labeling was initially directed appears to have contributed to per- only 30 percent of patients with sleep by the National Institute of Mental ceptions among clinicians that time difficulties seen in primary care clin- Health’s (NIMH’s) clinical guidelines, spent with patients is too short and ics had ever spoken with their physi- which have remained unchanged that patient load is excessive; these cians about a sleep problem; those since they were first published in 1984 were the findings from a survey of who did so reported that they––and (13). These early findings were based physician assistants, nurse practition- not their physicians––were the first to on data that suggested the risk of ers, and primary care physicians in a raise the subject of sleep difficulties. abuse and dependence associated large group model health mainte- Almost 36 percent of patients in the with benzodiazepines, as well as risks nance organization (7). Despite its chronic insomnia group reported that of tolerance, withdrawal, and rebound prevalence, insomnia is rarely dis- the physician’s recommendation for insomnia phenomena (14). It is im- cussed in visits to primary care physi- management was not effective. portant to note that these guidelines cians (8–10), and questions about Perception of treatments. There are now considered obsolete by the sleep may be swept aside in an effort is also a reticence among clinicians to National Institutes of Health (NIH). to save valuable office time. Physi- treat insomnia on the basis of their In referring to the 1984 guidelines, cians may avoid exploring problems misperceptions about the lack of effi- the NIH Web site now states, “This such as sleep difficulties in order to cacy of existing agents, restrictions on statement is outdated and is no longer

PSYCHIATRIC SERVICES ♦ http://ps.psychiatryonline.org ♦ March 2005 Vol. 56 No. 3 333 viewed by NIH as guidance for med- erbate episodes of mania (21). whether this hyperarousal leads to ical practice,” which highlights the It is important to note that correla- nonrestorative sleep. lack of guidance and direction for the tions between insomnia and comor- Therefore, it is not surprising that physician treating patients with in- bid illnesses, as well their outcomes, polysomnographic assessments of somnia (15). These concerns about do not necessarily imply causality. drugs used to treat insomnia, al- the risks of benzodiazepines have like- The lack of data demonstrating that though perhaps helpful for determin- ly persisted and have even been trans- insomnia has an impact on other ill- ing differences between drugs, may ferred to newer agents. nesses and that treating insomnia im- not reflect the patient’s subjective ex- Physicians may not be aware of be- proves outcomes in other illnesses is perience of those drugs, but the ideal havioral techniques for treating in- an inherent problem in the field of in- would be demonstrating efficacy by somnia, are generally not trained to somnia research, and more studies both polysomnography and patient perform them, and may not have suf- need to be conducted in this area. Al- report. It is thus important for the cli- ficient time with patients to provide though some physicians may have ob- nician to recognize that insomnia, as these treatments. Furthermore, be- served that treating secondary insom- defined by the DSM-IV, text revision cause of the paucity of therapists nia benefits patients, there is limited (DSM-IV-TR) (28), is a subjective trained in behavioral treatments for evidence that treating insomnia actu- clinical diagnosis, and therefore a pa- sleep disorders, most physicians can- ally improves outcomes in comorbid tient’s subjective report of sleep diffi- not easily refer patients for needed illness (22), and this may be another culties should direct management in therapy. factor in clinicians’ reluctance to rec- most cases. Questions about the Lack of evidence that treatment ognize or treat insomnia. range of symptoms experienced and improves outcomes. Although in- how they have altered over time are somnia is associated with substantial What is insomnia? important. Because insomnia is a pa- personal and societal impact, no The nature of insomnia itself proba- tient-reported symptom, rather than prospective studies have been done bly contributes to the difficulties as- a polysomnographically defined dis- to demonstrate that treating insomnia sociated with its treatment. order, referral to a sleep laboratory significantly improves outcomes of its Polysomnographic studies of patients for polysomnographic diagnosis associated comorbid conditions. with insomnia generally show abnor- should be reserved for cases in which Walsh and Ustun (16) estimated total malities such as prolonged latency to another primary sleep disorder, such direct annual cost in the United sleep onset, frequent arousals, and re- as obstructive sleep apnea or periodic States attributable to insomnia, in- duced amounts of total sleep. Howev- movement disorder, is suspected, be- cluding prescription and nonpre- er, objective measures of sleep do not cause these may require greater ex- scription medications, outpatient vis- always correlate well with the pa- pertise in sleep medicine. its to health care professionals, and tient’s experience of insomnia (23), The DSM-IV-TR defines insomnia inpatient or nursing home care, to be which may be partially due to the fact as “difficulty in initiating or main- approximately $12 billion for health that the function of sleep itself is still taining sleep or . . . nonrestorative care services and $2 billion for sleep- unknown, making it difficult to pin- sleep” and as “causing clinically sig- promoting agents. The impact of point which objective sleep abnor- nificant distress or impairment in so- treatment of insomnia on this sub- malities contribute to the clinical en- cial, occupational, or other important stantial financial burden is unknown, tity of insomnia. There is a tendency areas of functioning” (see the box on as is the impact of improving insom- to assume that insomnia is a problem the next page) (28). As such, all these nia on other negative correlates. of insufficient amount of sleep. Yet elements should be considered when People with insomnia report more epidemiologic studies consistently diagnosing and treating insomnia. days of limited activity, more days in demonstrate the lowest mortality The definition of sleep maintenance bed due to illness, greater health care rates for people who sleep an average remains fairly controversial, but it es- costs, and a higher incidence of mod- of seven hours per night, with in- sentially describes the patient’s ca- erate-to-severe occupational role dis- creasing risks for individuals who pacity to remain asleep throughout ability than people without insomnia sleep eight or more hours per night the night and reflects an assessment (17). Insomnia is also associated with (24). Furthermore, sleep deprivation of sleep quality. Arousals can be both greater health care utilization (17). does not mimic the effects of insom- objectively measured and subjective- Evidence also suggests that insom- nia; together, these data suggest that a ly reported. Problems with sleep nia may lead to the development of decreased quantity of sleep alone maintenance may consist of longer depression (18). Insomnia is also as- does not constitute insomnia (22). periods of arousal, which patients are sociated with an increased risk of sui- From a qualitative standpoint, how- generally aware of, or sleep fragmen- cide in depression (19) and resistance ever, data suggest that people with in- tation with brief EEG arousals (29), to cognitive-behavioral therapy (20). somnia show increased levels of which patients may not be aware of Finally, the strongest evidence for a arousal, evidenced by markers such as but that still result in the perception causal relationship between insomnia increased, higher-frequency elec- of nonrestorative sleep. Currently and subsequent illness is in bipolar troencephalographic (EEG) activity used metrics include wake time after disorder, where sleep loss has been during sleep and increased metabo- sleep onset, wake time during sleep demonstrated to precipitate or exac- lism (25–27), but it is not known onset, and number of awakenings.

334 PSYCHIATRIC SERVICES ♦ http://ps.psychiatryonline.org ♦ March 2005 Vol. 56 No. 3 Ideally, treatment of insomnia should address components of sleep onset, sleep maintenance, sleep DSM-IV, Text Revision criteria for primary insomnia quality, and improvement in next-day functioning. Difficulty initiating or maintaining sleep or nonrestorative sleep Causing clinically significant distress or impairment in social, occupational, or other important areas of functioning Symptom instability Not occurring exclusively during the course of another sleep disorder Although symptoms of insomnia have Not occurring exclusively during the course of a mental disorder been divided into categories of sleep Not due to the direct physiological effects of a substance or a general onset, sleep maintenance, and non- medical condition restorative sleep, symptom-specific classification may not always be useful clinically given that there is evidence of instability of insomnia symptoms over time for individual patients. A tion with insomnia have a diagnosable disorders (32,39) (Table 1). study of more than 2,500 general psychiatric disorder (10). Associations The best-studied comorbid psychi- practice clinic attendees demonstrat- between insomnia and psychiatric atric condition of insomnia is depres- ed that over a four-month period the disorders, particularly depression, are sion. Objective measures demon- nature of sleep complaints of sufferers even higher in clinical samples. Ap- strate loss of slow-wave (stage 3 and of insomnia changed, so that only half proximately three-quarters of pa- 4) sleep, frequent nocturnal awaken- of those who initially complained of tients with insomnia presenting to ings, and frequent arousals, all of sleep-onset problems continued to do sleep clinics or primary medical clin- which may lead to the perception of so. The remaining half developed new ics meet diagnostic criteria for psychi- nonrestorative sleep (32,39). Many of sleep symptoms, among them sleep- atric disorders (36). Overall, insomnia these symptoms are consistent with maintenance problems (30). is more strongly associated with de- sleep maintenance problems. There pression than it is with any other is also evidence of disturbed rapid eye Difficult-to-treat patients medical disorder in the primary care movement (REM) sleep architecture As noted, long-term prescription of setting (37). Symptoms of anxiety and in mood disorders, with a reduced la- medications for insomnia lacks an ev- depression were also strongly associ- tency from sleep onset to REM sleep idence base, has been discouraged by ated among children with insomnia in onset and an increased proportion of existing guidelines, and is a concern a pediatric clinic sample (38). REM sleep (39). Also, insomnia asso- among physicians (13). As shown be- Psychiatric patients present multi- ciated with psychiatric conditions fre- low, no medications are available that ple and varied sleep symptoms. Sleep quently lasts for the duration of the treat sleep maintenance symptoms EEG studies demonstrate significant illness, and, more important, sleep without the risk of next-day impair- decrements in sleep continuity—pro- symptoms have been found to persist ment. Chronic insomnia, including longed sleep latency, decreased sleep despite remission of depression (40). sleep maintenance problems, occurs efficiency, and decreased total sleep Medical disorders. Individuals more commonly among the elderly across the night––in most groups of with insomnia have higher rates of (31), depressed patients (32), and psychiatric patients compared with medical illnesses than those without medically ill populations (33,34), in- control patients without psychiatric sleep problems. Mellinger and col- cluding those with chronic pain syn- dromes (34). These patients are often viewed as difficult to treat yet are Table 1 among the groups that have the great- Sleep characteristics of patients with psychiatric disorders versus control patientsa est need for treatment. Psychiatric disorders. Insomnia Sleep Percentage of Percentage of b is a very common feature of psychi- Disorder continuity slow-wave sleep REM latency REM sleep atric disorders (33). Prevalence rates Affective Consistently Consistently Consistently Consistently of insomnia are greatly increased disorder decreased decreased decreased increased among persons with psychiatric ill- Anxiety Consistently No change No change No change nesses. A European epidemiologic decreased sample of 1,536 people found that Schizophrenia Consistently No change Decreased in No change decreased some studies significant insomnia was present in 71 Eating disorder Decreased in No change Decreased in No change percent, 69 percent, 61 percent, and some studies some studies 32 percent of those with dementia, Alcoholism Consistently Consistently No change Increased in depressive disorders, anxiety disor- decreased decreased some studies ders, and alcoholism, respectively Insomnia Consistently Consistently No change No change decreased decreased (35). Furthermore, on the basis of epidemiologic studies, up to 40 per- a Data from meta-analysis by Benca and associates (39) cent of adults in the general popula- b Rapid eye movement

PSYCHIATRIC SERVICES ♦ http://ps.psychiatryonline.org ♦ March 2005 Vol. 56 No. 3 335 leagues (33) determined that 53 per- long-lasting changes in chronic insom- objective sleep onset latency after ter- cent of adults with serious insomnia nia sufferers for several sleep parame- mination of treatment, either alone or had two or more health problems, ters. Seventy to 80 percent of patients combined with pharmacotherapy fol- compared with only 24 percent of treated with nonpharmacologic inter- lowed by combination therapy, those with no sleep difficulties. In ad- ventions were found to benefit from whereas effects of pharmacotherapy dition, Ford and Kamerow’s study treatment; however, there was signifi- persisted only during acute adminis- (10) found that patients with insom- cant variability in the magnitude of tration. However, subjectively report- nia used general medical services treatment response. Stimulus control, ed total sleep time showed similar in- more frequently, and higher rates of progressive muscle relaxation, and creases after termination of treatment insomnia have been documented paradoxical intention all met the in both the CBT and zolpidem among primary care patients than in American Psychological Association’s groups; the second largest increase the general population (9). There is criteria for empirically supported psy- was in the combination therapy group. also evidence that greater severity of chological treatments for insomnia, No between-group differences were sleep disturbance is correlated with and sleep restriction, biofeedback, found for objective measures. Al- worse outcomes among patients who and multifaceted cognitive-behavioral though these data support the superi- experience pain (34), that among pa- therapy (CBT) were viewed as “prob- or efficacy of behavioral treatment tients with pain insomnia is correlated ably efficacious” treatments (47). over pharmacotherapy, interpretation with the development of depression Evidence for favoring pharmaco- may be limited by the small samples, (34), and that the presence of both logic therapy over behavioral therapy the absence of sleep maintenance in- depression and insomnia contributes or vice versa was inconclusive (48). somnia or comorbid conditions among to highest pain severity (41). Comparisons of hypnotic and behav- these patients, and the failure to as- Insomnia is also correlated with ioral therapies suggested that hypnot- sess the effects of treatment on day- worse outcomes in a number of med- ic drugs produce more rapid improve- time function. More research in larg- ical illnesses, including an increased ments in sleep in comparison with be- er patient populations is needed to risk of mortality among institutional- havioral treatments such as relaxation clarify the relative efficacies of the ized elderly people (42), greater dis- and sleep hygiene education (48) and different treatments for insomnia. ability among stroke patients (43), and that treatment effects over longer pe- Clearly, efficacy coupled with mini- increased risk of mortality among pa- riods (four to eight weeks) are similar mal side effects makes behavioral tients with cardiovascular disease (44). for pharmacologic and nonpharmaco- techniques highly recommended for Elderly patients. Older patients logic therapies, as well as the combi- treating insomnia; however, factors often complain that their sleep is non- nation of the two (49). On the other such as cost, lack of availability, and restorative and that they have difficul- hand, longer-term outcome studies, potential problems with patient moti- ty staying asleep (31). Foley and col- such as those of six to 24 months, have vation and compliance may make the leagues (2) found that 30 percent of indicated that clinical benefit is not as use of behavioral techniques difficult. elderly patients complained of awak- well maintained over time after dis- Optimal patient- and treatment-relat- ening during the night, 19 percent continuation of pharmacologic treat- ed variables have not been clearly de- complained of waking up too early, ment as it is after cessation of behav- fined (47). A meta-analysis of 59 stud- and 19 percent said they had trouble ioral therapy (48,49). Patients who ies assessing nonpharmacologic treat- falling asleep. Polysomnographic find- have received combined treatments ment of chronic insomnia found that ings have also suggested that the pri- do not appear to have as good a long- an average of five hours of therapy mary change in the sleep of older term outcome as those who receive was provided (52). The results sug- adults is an inability to sustain sleep behavioral therapy alone (50), al- gested that stimulus control and sleep through the night (45). These findings though reasons for this finding have restriction techniques were most also suggest that daytime sleepiness not been clearly elucidated (47). helpful in producing improvements among healthy elderly patients with- A recent randomized controlled over an average six-month follow-up out insomnia does not correlate with study compared CBT, zolpidem, and period, but sleep hygiene treatment total sleep time or any sleep stage but the combination of the two therapies alone was not deemed effective. is significantly correlated with meas- among 63 patients with sleep-onset Most studies have been conducted ures of sleep fragmentation (46). insomnia and impaired daytime func- with highly trained therapists such as tioning (51). Participants received psychologists, who may not be avail- Nonpharmacologic four weeks of active treatment with able to a primary care physician’s of- management approaches 10 mg of zolpidem, followed by 5 mg fice, and evidence indicates that a An American Academy of Sleep Med- of zolpidem nightly for one week, and lower level of training––a trainee icine task force reviewed 48 clinical then 5 mg of zolpidem every second therapist, for example––is associated trials and two meta-analyses in an at- night for one week. CBT of 30 min- with worse outcomes (47). However, tempt to develop practice guidelines utes per each session was offered several recent studies have suggested for nondrug alternatives for managing weekly for three weeks, with the final that more limited interventions may primary chronic insomnia (47). Its session two weeks later. CBT was the be helpful to patients with insomnia. findings indicated that nonpharmaco- most effective intervention in this CBT was equally helpful when ad- logic therapies produce reliable and study in reducing both subjective and ministered in individual, face-to-face

336 PSYCHIATRIC SERVICES ♦ http://ps.psychiatryonline.org ♦ March 2005 Vol. 56 No. 3 treatment, in group therapy sessions, or through brief telephone contact (53). In another study, two sessions of Sleep hygiene rules CBT led to greater improvement for people with chronic primary insom- Wake up at the same time every day, regardless of when you went to sleep. Maintain a consistent bedtime. nia than two sessions of generic sleep Exercise regularly, preferably in the late afternoon, but not within two to hygiene education (54). More re- four hours of bedtime. search is required to determine the Perform relaxing activities before bed. optimal behavioral treatment inter- Keep your bedroom quiet and cool (extreme temperatures compromise ventions for insomnia in the primary sleep). Do not watch the clock at night. care setting. Avoid caffeine and nicotine for at least six hours before bedtime. Although sleep hygiene techniques Drink alcohol only in moderation and avoid use for at least four hours before alone do not necessarily lead to con- bedtime. sistent improvement in insomnia Avoid napping; it may interfere with the ability to fall asleep at night. (55), they are usually a part of most behavioral treatments for insomnia. Sleep hygiene recommendations are listed in the box on this page, and treatments recommended by the zaleplon has resulted in more effec- used benzodiazepine hypnotic (74). American Academy of Sleep Medi- tive sleep-onset agents with minimal Objective sleep laboratory data re- cine for the nonpharmacological next-day residual effects but possibly garding the ability of temazepam to treatment of insomnia are described with decreased efficacy as sleep- improve sleep maintenance as defined in Table 2. maintaining agents (61–64). by number of awakenings during sleep In addition, no currently approved have been equivocal. Two objective Pharmacologic insomnia agents have been evaluated studies did not find significant reduc- treatment approaches for treating chronic insomnia, as evi- tions in number of awakenings with As discussed above, insomnia symp- denced by the fact that randomized doses of 15 mg to 30 mg of temazepam toms tend to change over time. In ad- controlled studies of zolpidem, zale- (72,75). Another study (76) of 48 peo- dition, a survey of a general popula- plon, and approved benzodiazepines ple with insomnia did show a signifi- tion sample has shown that a majority have not exceeded four weeks for za- cant improvement compared with of patients with both transient and leplon (62,64), five weeks for zolpi- placebo for 15 mg and 30 mg of chronic insomnia have sleep-mainte- dem (63), or 12 weeks for temazepam temazepam on number of awakenings, nance problems (56). In recent years (65) of continuous use. Although it is although significant reductions were attempts have been made to charac- currently unknown what duration of seen at only one of the two sites used terize the ideal hypnotic agent continued efficacy would need to be for the study. Temazepam may also be (57,58). Attributes proposed for an demonstrated in order to support associated with the development of ideal agent include rapid onset of ac- long-term use of an agent for treating tolerance (77). Although adverse tion and elimination; improved ability insomnia, there is a clear need to events associated with use of to initiate and maintain sleep; im- demonstrate efficacy of medications temazepam have not been explored to proved sleep quality; normal sleep ar- over longer periods. the same extent as those of other ben- chitecture; improved daytime per- Benzodiazepines. Currently ap- zodiazepines, benzodiazepine use in formance and well-being with mini- proved benzodiazepines for the treat- general has been associated with day- mal drug interactions; the absence of ment of insomnia are flurazepam, tri- time drowsiness (71), memory impair- hangover effects or unwanted side ef- azolam, quazepam, estazolam, and ment (71), psychomotor impairment fects; and no significant potential for temazepam. Both subjective and ob- (71), and risk of tolerance, abuse, and tolerance, abuse, dependence, with- jective studies have generally found dependence (72). Temazepam itself drawal, or rebound effects. improvements in sleep maintenance has been implicated in causing next- Until the advent of the nonbenzo- measures, specifically wake time after day sedation (78) and impairment in diazepine hypnotics, the most com- sleep onset and number of awaken- memory and cognition (78) and has monly used agents were benzodi- ings, with the longer-acting agents demonstrated evidence of rebound in- azepines. Aside from triazolam, most like flurazepam, quazepam, and esta- somnia on withdrawal and depend- benzodiazepine hypnotics, which in- zolam (66–70). Their use, however, is ence liability (72). Triazolam has also clude flurazepam, estazolam, quaze- also associated with next-day sedation been implicated in causing rebound pam, and temazepam, have long half- and impaired cognitive and psy- insomnia (79). lives, which contribute to their effica- chomotor function (71). Triazolam’s Currently available nonbenzo- cy in maintaining sleep throughout efficacy with regard to measures of diazepine hypnotics. Evidence for the night but also may result in next- sleep maintenance is less evident, the utility of currently available non- day impairments (59,60). The advent probably because of its shorter half- benzodiazepine hypnotics points to of shorter-acting triazolam and the life (72,73). their primary efficacy as sleep-onset, nonbenzodiazepines zolpidem and Temazepam is the most commonly rather than as sleep-maintenance,

PSYCHIATRIC SERVICES ♦ http://ps.psychiatryonline.org ♦ March 2005 Vol. 56 No. 3 337 Table 2 Recommended behavioral therapies for insomnia

Level of recommendation Therapy Description

Empirically supported Stimulus control The main objective is to reassociate the bed and bedroom with the treatmentsa rapid onset of sleep. Instructions: Go to bed only when sleepy; use the bed and bedroom only for sleep; leave bed and go into another room when unable to fall asleep or return to sleep easily; return to bed only when sleepy again; maintain a regular morning rising time regardless of duration of sleep the previous night; avoid daytime naps. Progressive muscle relaxation This method is based on the idea that mental relaxation will be a natural outcome of physical relaxation—it is a deep-relaxation technique. Instructions: Tense or tighten one muscle group at a time, then release tension; muscle groups are tightened and relaxed one at a time in a specific order; a greater degree of muscle tension is at- tempted in subsequent exercises as patient becomes familiar with the technique. Paradoxical intention Based on the concept that performance anxiety contributes to preventing proper sleep: persuade the patient to engage in the most feared behavior (that is, staying awake); as the patient stops try- ing to fall asleep, the performance anxiety related to attempting to fall asleep is reduced. Probably efficacious Sleep restriction The objective is to reduce time in bed to lower the chance of frag- treatmentsb mented and poor-quality sleep. Instructions: Reduce amount of time spent in bed to increase the percentage of time spent asleep—improves patient’s sleep efficiency (time asleep/time in bed); time allowed in bed per night is increased by 15 to 30 minutes as sleep efficiency improves; adjust- ments are made over a period of weeks until optimal sleep duration is achieved—best to alter bedtime and maintain constant rising time to maintain a regular sleep-wake rhythm; to minimize daytime sleepiness, time in bed should not be reduced to fewer than five hours per night. Creates a mild state of sleep deprivation, to pro- mote more rapid sleep onset and more efficient sleep. Biofeedback Therapeutic technique that teaches patients how to facilitate in- creased slow brain wave activity (and thus facilitate falling asleep) by using electroencephalographic (EEG) monitoring. Eventually the patient is able to apply this skill without the use of the EEG. Multifaceted cognitive The goal is to identify dysfunctional beliefs and attitudes about behavioral therapy sleep and replace them with more adaptive substitutes. Treatment targets: Unrealistic sleep expectations (“I must get eight hours of sleep per night”); misconceptions regarding causes of in- somnia (“My insomnia is due to a chemical imbalance”); amplifica- tion of consequences of insomnia (“I can do nothing after a bad night’s sleep”); performance anxiety due to excessive attempts at controlling the sleep process. a According to the American Psychiatric Association b Supported by the American Psychiatric Association

agents. Once again, longer-term ran- there has been less evidence of sub- with zaleplon and zolpidem than with domized, double-blind, controlled jective and objective next-day resid- benzodiazepines (84), which may be studies that demonstrate efficacy of ual effects associated with zolpidem related to the differences in CYP me- these agents have not been per- (4,63,82) or subjective next-day im- tabolism. Whereas triazolam, for ex- formed, but safety over the longer pairment with zaleplon, even when ample, is biotransformed almost en- term has been demonstrated in open- the latter has been delivered in the tirely via CYP3A4, the newer agents label studies (80,81), with minimal middle of the night (83). are biotransformed by CYP3A4 and evidence of rebound phenomena. By Fewer clinically important drug- several other CYP isozymes, which comparison with benzodiazepines, drug interactions appear to occur means that CYP3A4 inhibitors and in-

338 PSYCHIATRIC SERVICES ♦ http://ps.psychiatryonline.org ♦ March 2005 Vol. 56 No. 3 ducers may have less effect on their (89). The relationship between hyp- commonly prescribed agents for in- biotransformation. On the other notic use and falls is complicated by somnia (the other being zolpidem) hand, because these agents are new- the fact that sleep problems among (74). Trazodone’s precise mecha- er, only a few studies have been con- elderly people are independently as- nisms of action have not been deter- ducted (84), and further research is sociated with an increased risk of falls mined; it is believed to be a weak but needed. (90). Hallucinatory phenomena and specific inhibitor of synaptosomal re- Zolpidem. Zolpidem is an effec- other sensory distortions have been uptake of serotonin, and its thera- tive sleep-onset agent and has consis- reported even with therapeutic doses peutic effects may also be based on tently demonstrated reduced time to of zolpidem (91). Possibly because of the serotonergic 5-HT1a, 5-HT1c, sleep onset. It was the most com- its limited efficacy for sleep mainte- and 5-H2 receptors (93). As an anti- monly prescribed agent for insomnia nance problems, patients may take depressant, trazodone has the advan- in 2001 (74) despite the absence of higher than recommended doses or tages of low cost, no restrictions on studies of nightly use for longer than take a second dose during the night, long-term prescription, and a low five weeks (63). One (63) of the two which may increase the risk of both abuse potential compared with ben- (63,85) objective studies of zolpidem acute side effects and next-day resid- zodiazepine receptor agonists (94). efficacy in primary insomnia demon- ual effects (83). However, the risk of side effects as- strated significantly improved sleep Zaleplon. Zaleplon is a nonbenzo- sociated with trazodone is not trivial efficiency; however, neither demon- diazepine of the and includes orthostatic hypotension strated improvement over placebo in class with a very short elimination and blurred vision (95), which in- reducing wake time after sleep onset half-life of about one hour (62). Sev- crease the risk of falls, especially or number of awakenings (63,85). eral studies have demonstrated that among elderly patients. Symptoms Another study, conducted with 15 zaleplon is effective in reducing la- such as nausea, dry mouth, constipa- patients with nonorganic insomnia tency-to-sleep-onset insomnia in ran- tion, drowsiness, headache, and oth- related to neurotic or stress-related domized, controlled, double-blind er central nervous system effects also disorders, indicated that, although subjective studies (62,64). It is asso- increase morbidity (95,96). Priapism total sleep time improved signifi- ciated with a minimum of next-day is a rare side effect of trazodone, not cantly with 10 mg of zolpidem versus side effects or residual sedation, associated with use of benzodi- placebo, there were no statistically making it a useful agent for sleep-on- azepine receptor agonists, and its oc- significant differences in wake time set problems (92). However, there currence is considered a urological during the sleep period or number of are minimal data to support zale- emergency (97). awakenings (86). These studies plon’s sleep-maintenance efficacy Given these drawbacks, it is im- (63,86) suggest that zolpidem’s effi- throughout the night when given at portant to ask how effective tra- cacy in improving sleep efficiency sleep onset. In double-blind, place- zodone is with respect to insomnia in may be related more to its efficacy as bo-controlled studies, doses of less general and sleep maintenance in a sleep onset agent rather than a than 20 mg did not result in signifi- particular. Surprisingly, considering sleep maintenance agent. Subjective cant increases in subjectively as- the drug’s popularity, there is a studies of zolpidem versus placebo sessed total sleep time (62,64). paucity of data supporting its use for have not consistently demonstrated Whereas most clinically significant insomnia, especially in primary in- significant improvement in sleep changes, such as reduced sleep laten- somnia populations, because tra- maintenance parameters, such as cy and increased sleep duration and zodone studies have usually been wake time after sleep onset and sleep quality, were seen at the 20 mg conducted among depressed pa- number of awakenings (4,82). dose in studies by Elie and associates tients (98) and have utilized small Although next-day benefits with (62) and Fry and colleagues (64), the samples (N≤22) (98). Objective zolpidem use have not been clearly recommended dose of 10 mg did not studies have not exceeded eight evaluated or demonstrated (86,87), a result in increased sleep duration. weeks (99). Two (3,98) of the three study by Saletu-Zyhlarz (86) indicat- Only in the Fry study (64) did 20 mg studies (3,98,100) that objectively ed that there was significant improve- of zaleplon improve number of awak- measured waking time after sleep ment in somatic complaints versus enings (wake time after sleep onset onset did not show benefit. This placebo. All other tests of psychomo- was not measured), whereas the 10 finding was also reflected in studies tor function, attention, and memory, mg dose had no effect on number of that used subjective measures (101, as well as subjective reports of well- awakenings, suggesting that higher 102). Although trazodone showed being, showed no difference com- dosing may be necessary for some pa- statistically significant subjective im- pared with placebo. One retrospec- tients. At present, however, little is provements from baseline in ease of tive case-control study found that use known about side effects associated falling asleep and quality of sleep, it of zolpidem by older people was asso- with use of the 20 mg dose. was also associated with negative ef- ciated with nearly twice the risk of hip Trazodone. Despite limited evi- fects in terms of ease of awakening fracture (88), although evidence gen- dence for its efficacy, questions and feelings at or after wakening, erally points to the fact that longer- about its side effect profile, and no relative to baseline. This finding sug- acting agents are more likely to be as- FDA approval for use as a hypnotic, gests that the improvements in sleep sociated with falls and hip fractures trazodone is one of the two most onset and quality may be counterbal-

PSYCHIATRIC SERVICES ♦ http://ps.psychiatryonline.org ♦ March 2005 Vol. 56 No. 3 339 anced by negative next-day effects. Evidence-based research and this study, after the first week and at Studies have also demonstrated evi- the difficult-to-treat patient the end of six months of treatment, 3 dence of declining efficacy after one Little work has been done to evaluate mg of eszopiclone significantly re- to two weeks of treatment, suggest- the use of hypnotic agents for patients duced time to sleep onset and im- ing the possible development of tol- with comorbid psychiatric or medical proved measures of sleep mainte- erance (3,4). Trazodone has also conditions, although one study nance, including wake time after been associated with rebound in- demonstrated improvement in total sleep onset and number of awaken- somnia after withdrawal (3). sleep time, daytime sleepiness, and ings. Sleep quality was also im- Over-the-counter medications. morning stiffness among patients proved. Furthermore, patients re- Sedating antihistamines are fre- with rheumatoid arthritis who were ported significant subjective im- quently used as sleep aids (103), treated with triazolam (112). provements in daytime alertness, possibly because of their perceived Only one study has evaluated ad- sense of well-being, and daytime safety and low cost. A 2002 Nation- junctive hypnotic therapy for patients ability to function. No evidence was al Sleep Foundation poll of 1,000 successfully treated for depression found of clinically significant toler- people found that 24 percent of who had residual insomnia. Zolpi- ance or residual, next-day effects. those who reported sleep difficul- dem as an adjunct to a selective sero- This study is the first long-term, ties used over-the-counter or store- tonin reuptake inhibitor was associat- placebo-controlled evaluation that bought sleep aids; 5 percent of ed with longer sleep times, greater was conducted with any hypnotic those who used over-the-counter sleep quality, and reduced number of that evaluated the effect of treatment drugs reported use every night or a awakenings than was a selective sero- on both sleep and next-day function- few nights a week (104). Diphenhy- tonin reuptake inhibitor and placebo ing, and it is notable that the findings dramine, the most widely used over- over four weeks (113). As noted, tra- demonstrated significant improve- the-counter antihistamine sleep aid, zodone has been investigated as an ments in all four components of the appears to be superior to placebo in adjunctive therapy in antidepressant- DSM-IV-TR definition of primary in- several double-blind, placebo-con- induced insomnia and was shown to somnia. Polysomnographic data were trolled studies (105,106), although decrease the number of nocturnal not obtained, however, and longer- neither of these studies used objec- awakenings in a one-week study term conclusions about drug-seeking tive polysomnographic measure- (114); it improved overall sleep bet- behavior and discontinuation effects ments. Most studies were of very ter than placebo in another small could not be drawn from this study short duration and involved patients study (115). There is also evidence (121). The six-month findings were with mild-to-moderate insomnia. that among patients with bipolar dis- replicated in another six-week, dou- No recent controlled studies order, manic episodes resolve more ble-blind, randomized, placebo-con- demonstrate efficacy of diphenhy- quickly for patients who sleep more trolled, polysomnographic study, in dramine for longer than three (116), although randomized, con- which efficacy was demonstrated by weeks for objectively determined trolled trials assessing hypnotic ther- using both subjective and objective measures of sleep maintenance. apy are yet to be conducted. Evi- sleep measures, which were well cor- Diphenhydramine also appears to dence shows that behavioral therapy related (122). Eszopiclone received produce tolerance to its sleep-in- is as effective for patients with sec- FDA approval for sleep onset and ducing effects within a few days ondary insomnia as it is for those with sleep maintenance insomnia in late (107). Moreover, over-the-counter primary insomnia (117–119). 2004. antihistamines have substantial Another new agent is indiplon, a neurocognitive effects, including Discussion nonbenzodiazepine GABAA modu- next-day sedation (105) and im- Recent developments in pharmaco- lator. In a-yet-to-be-published study paired psychomotor and cognitive logic therapy have yielded promising of elderly patients (123), modified- function (108,109). new agents that may be effective for release indiplon, at doses of 20 mg, The potential for diphenhydra- sleep maintenance, as evidenced in 30 mg, and 35 mg, was reported to mine-related toxicity and drug-drug clinical trials of patients with insom- significantly improve objective poly- interactions is substantial (110). Side nia. Eszopiclone is a novel nonbenzo- somnographic measures of sleep effects include urinary retention and diazepine (cyclopyrrolone) agent maintenance (wake time after sleep blurred vision (111); orthostatic hy- with a half-life of five to six hours onset and number of awakenings) potension, dizziness, and palpitations (120). A recently published study and sleep-onset latency compared (105,111); increased liver enzymes demonstrated that 3 mg of eszopi- with placebo over two nights of ac- (105,106); and drowsiness, dizziness, clone was effective in improving pa- tive treatment in a group of patients grogginess, and tiredness (105). The tient-reported sleep and next-day selected for their sleep maintenance available evidence suggests that function in patients with chronic in- difficulties (elevated amount of wak- diphenhydramine and related over- somnia (121). The study was a ran- ing time after sleep onset). Subjec- the-counter antihistamines do not domized, double-blind, placebo-con- tive measures of sleep were also sig- represent a viable treatment strategy trolled design evaluating nightly nificantly improved. Because the for long-term sleep maintenance in treatment for six months among 788 second eszopiclone and indiplon chronic insomnia. patients with chronic insomnia. In studies have been reported only in

340 PSYCHIATRIC SERVICES ♦ http://ps.psychiatryonline.org ♦ March 2005 Vol. 56 No. 3 abstract form and have not yet been ceuticals, -Aventis, Sepracor, Inc., 18. Breslau N, Roth T, Rosenthal L, et al: Sleep published, it is difficult to assess these Takeda Pharmaceuticals, and . disturbance and psychiatric disorders: a longitudinal epidemiological study of studies fully. young adults. Biological Psychiatry References 39:411–418, 1996 Conclusions 1. Roth T: New developments for treating 19. Agargun MY, Kara H, Solmaz M: Sleep dis- There are numerous barriers to the sleep disorders. Journal of Clinical Psychia- turbances and suicidal behavior in patients try 62(suppl):10:3–4, 2001 appropriate recognition, diagnosis, with major depression. Journal of Clinical Psychiatry 58:249–251, 1997 and treatment of insomnia, despite 2. Foley DJ, Monjan AA, Brown SL, et al: Sleep complaints among elderly persons: the fact that it is highly prevalent 20. Thase ME, Simons AD, Reynolds CF, III: an epidemiologic study of three communi- Abnormal electroencephalographic sleep and associated with a number of ad- ties. Sleep 18:425–432, 1995 profiles in major depression: association verse personal and socioeconomic 3. Montgomery I, Oswald I, Morgan K, et al: with response to cognitive behavior thera- consequences. Trazodone enhances sleep in subjective py. Archives of General Psychiatry 53:99– Behavioral therapies, which are at quality but not in objective duration. 108, 1996 British Journal of Clinical Pharmacology 21. Wehr TA: Sleep loss: a preventable cause of least as effective as pharmacologic 16:139–144, 1983 therapies and appear to have longer- mania and other excited states. Journal of 4. Walsh JK, Erman M, Erwin CW, et al: Sub- Clinical Psychiatry 50(suppl):8–16, 1989 lasting efficacy after cessation, should jective hypnotic efficacy of trazodone and always be offered to patients com- 22. Benca RM: Consequences of insomnia and zolpidem in DSM-III-R primary insomnia. its therapies. Journal of Clinical Psychiatry plaining of insomnia, either alone or Human Psychopharmacology 13:191–198, 62(suppl 10):33–38, 2001 in combination with pharmacologic 1998 23. Erman MK: Sleep architecture and its rela- 5. Rosen RC, Rosekind M, Rosevear C, et al: therapy. tionship to insomnia. Journal of Clinical Physician education in sleep and sleep dis- Psychiatry 62(suppl 10):9–17, 2001 Commonly prescribed newer hyp- orders: a national survey of US medical notic medications demonstrate schools. Sleep 16:249–254, 1993 24. Youngstedt SD, Kripke DF: Long sleep and sleep-onset efficacy but limited mortality: rationale for sleep restriction. 6. Papp KK, Penrod CE, Strohl KP: Knowl- Sleep Medicine Reviews 8:159–174, 2004 sleep-maintenance efficacy, and old- edge and attitudes of primary care physi- er agents, although effective at main- cians toward sleep and sleep disorders. 25. Perlis ML, Merica H, Smith MT, et al: Beta taining sleep, may produce substan- Sleep and Breathing 6:103–109, 2002 EEG activity and insomnia. Sleep Medi- cine Reviews 5:363–374, 2001 tial next-day residual effects. In ad- 7. Freeborn DK, Hooker RS, Pope CR: Satis- dition, data for efficacy and safety of faction and well-being of primary care 26. Krystal AD, Edinger JD, Wohlgemuth WK, providers in managed care. Evaluation and et al: NREM sleep EEG frequency spectral any agents in long-term use are lim- the Health Professions 25:239–254, 2002 correlates of sleep complaints in primary ited. Trazodone is also commonly insomnia subtypes. Sleep 25:630–640, 2002 8. Hajak G: Insomnia in primary care. Sleep prescribed as a sleep agent; however, 23(suppl):3:S54–S63, 2000 27. Bonnet MH, Arand DL: Insomnia, meta- evidence for its efficacy is limited, 9. Shochat T, Umphress J, Israel AG, et al: In- bolic rate, and sleep restoration. Journal of and its next-day effects and side ef- somnia in primary care patients. Sleep Internal Medicine 254:23–31, 2003 fects are significant. 22(suppl):2:S359–S365, 1999 28. American Psychiatric Association: Diagnos- Patients who suffer from longer- 10. Ford DE, Kamerow DB: Epidemiologic tic and Statistical Manual of Mental Disor- term insomnia and have predomi- study of sleep disturbances and psychiatric ders–Text revision. 4th ed. Washington, DC, American Psychiatric Publishing, 2000 nantly sleep-maintenance problems, disorders. An opportunity for prevention? JAMA 262:1479–1484, 1989 such as those with psychiatric or 29. Stepanski EJ: The effect of sleep fragmen- 11. Nowell PD, Mazumdar S, Buysse DJ, et al: tation on daytime function. Sleep 25:268– medical illnesses, are particularly 276, 2002 challenging for clinicians, because Benzodiazepines and zolpidem for chronic insomnia: a meta-analysis of treatment effi- 30. Hohagen F, Kappler C, Schramm E, et al: little research has been conducted in cacy. JAMA 278:2170–2177, 1997 Sleep onset insomnia, sleep maintaining in- these patient populations with either 12. Physicians’ Desk Reference: 57th ed. somnia, and insomnia with early morning behavioral therapy or pharmacother- Montvale, NJ, Medical Economics, 2003 awakening: temporal stability of subtypes in apy, and guidelines are lacking. Al- a longitudinal study on general practice at- 13. National Institutes of Health: Consensus tenders. Sleep 17:551–554, 1994 though a recently completed six- conference. Drugs and insomnia: The use month study has demonstrated that of medications to promote sleep. JAMA 31. Reynolds CF, III: Sleep disorders, in Com- 251:2410–2414, 1984 prehensive Review of Geriatric Psychia- newer hypnotic agents can have long- try–II, 2nd ed. Edited by Sadavoy J, term efficacy and a positive impact 14. Greenblatt DJ, Shader RI, Abernethy DR: Lazarus LW, Jarvik LF, et al. Washington, on subjective next-day functioning, Drug therapy. Current status of benzodi- DC, American Psychiatric Press, 1996 azepines. New England Journal of Medi- more research is needed to improve cine 309:354–358, 1983 32. Szuba MP, Fernando AT, Groh-Szuba G: the therapeutic armamentarium for Sleep abnormalities in treatment-resistant 15. Drugs and insomnia: the use of medica- mood disorders, in Treatment Resistant difficult-to-treat patients as well as to tions to promote sleep. NIH Consensus Mood Disorders. Edited by Amsterdam determine the long-term benefits of statement, 1983. Available at http://consen- JD, Nierenberg AA. Cambridge, UK, Cam- treating insomnia. ♦ sus.nih.gov/cons/039/039_intro.htm bridge University Press, 2001 16. Walsh J, Ustun B: Prevalence and health 33. Mellinger GD, Balter MB, Uhlenhuth EH: Acknowledgments consequences of insomnia. Sleep 22: Insomnia and its treatment. Prevalence and S427–S436, 1999 correlates. Archives of General Psychiatry The author thanks Dr. Jacqui Brooks and 42:225–232, 1985 Sepracor, Inc. for their assistance in the 17. Simon GE, VonKorff M: Prevalence, bur- den, and treatment of insomnia in primary 34. McCracken LM, Iverson GL: Disrupted preparation of this paper. The author is a care. American Journal of Psychiatry 154: sleep patterns and daily functioning in pa- consultant or speaker for King Pharma- 1417–1423, 1997 tients with chronic pain. Pain Research and

PSYCHIATRIC SERVICES ♦ http://ps.psychiatryonline.org ♦ March 2005 Vol. 56 No. 3 341 Management 7:75–79, 2002 51. Jacobs GD, Pace-Schott EF, Stickgold R, et 66. Scharf MB, Roth PB, Dominguez RA, et al: al: Cognitive behavior therapy and pharma- Estazolam and flurazepam: a multicenter, 35. Weyerer S, Dilling H: Prevalence and treat- cotherapy for insomnia. Archives of Inter- placebo-controlled comparative study in ment of insomnia in the community: results nal Medicine 164:1888–1896, 2004 outpatients with insomnia. Journal of Clini- from the Upper Bavarian Field Study. cal Pharmacology 30:461–467, 1990 Sleep 14:392–398, 1991 52. Morin CM, Culbert JP, Schwartz SM: Non- pharmacological interventions for insom- 67. Melo de Paula AJ: Comparative study of 36. Buysse DJ, Reynolds CF, III, Hauri PJ, et nia: a meta-analysis of treatment efficacy. and flurazepam in the treat- al: Diagnostic concordance for DSM-IV American Journal of Psychiatry 151:1172– ment of insomnia. Clinical Therapeutics 6: sleep disorders: a report from the 1180, 1994 500–508, 1984 APA/NIMH DSM-IV field trial. American Journal of Psychiatry 151:1351–1360, 1994 53. Bastien CH, Morin CM, Ouellet MC, et al: 68. Aden GC, Thatcher C: Quazepam in the Cognitive-behavioral therapy for insomnia: short-term treatment of insomnia in outpa- 37. Katz DA, McHorney CA: The relationship comparison of individual therapy, group tients. Journal of Clinical Psychiatry 44: between insomnia and health-related qual- therapy, and telephone consultations. Jour- 454–456, 1983 ity of life in patients with chronic illness. nal of Consulting and Clinical Psychology Journal of Family Practice 51:229–235, 72:653–659, 2004 69. Hernandez LR, Del Rosal PL, Ponce MC: 2002 Short-term study of quazepam 15 mil- 54. Edinger JD, Sampson WS: A primary care ligrams in the treatment of insomnia. The 38. Stein MA, Mendelsohn J, Obermeyer WH, “friendly” cognitive behavioral insomnia Journal of International Medical Research et al: Sleep and behavior problems in therapy. Sleep 26:177–182, 2003 11:162–166, 1983 school-aged children. Pediatrics 107:E60, 2001 55. Stepanski EJ, Wyatt JK: Use of sleep hy- 70. Cohn JB, Wilcox CS, Bremner J, et al: Hyp- giene in the treatment of insomnia. Sleep notic efficacy of estazolam compared with 39. Benca RM, Obermeyer WH, Thisted RA, Medicine Reviews 7:215–225, 2003 flurazepam in outpatients with insomnia. et al: Sleep and psychiatric disorders: a Journal of Clinical Pharmacology 31:747– meta-analysis. Archives of General Psychia- 56. Ancoli-Israel S, Roth T: Characteristics of 750, 1991 try 49:651–668, 1992 insomnia in the United States: results of the 1991 National Sleep Foundation survey: I. 71. Holbrook AM, Crowther R, Lotter A, et al: 40. Reynolds CF, III, Hoch CC, Buysse DJ, et Sleep 22(suppl 2):S347–S353, 1999 Meta-analysis of benzodiazepine use in the al: Sleep in late-life recurrent depression: treatment of insomnia. Canadian Medical changes during early continuation therapy 57. Weitzel KW, Wickman JM, Augustin SG, et Association Journal 162:225–233, 2000 with nortriptyline. Neuropsychopharma- al: Zaleplon: a pyrazolopyrimidine sedative- cology 5:85–96, 1991 hypnotic agent for the treatment of insom- 72. Kales A, Manfredi RL, Vgontzas AN, et al: nia. Clinical Therapeutics 22:1254–1267, Rebound insomnia after only brief and in- 41. Wilson KG, Eriksson MY, D’Eon JL, et al: 2000 termittent use of rapidly eliminated benzo- Major depression and insomnia in chronic diazepines. Clinical Pharmacology and pain. Clinical Journal of Pain 18:77–83, 58. Morin CM: Measuring outcomes in ran- Therapeutics 49:468–476, 1991 2002 domized clinical trials of insomnia treat- ments. Sleep Medicine Reviews 7:263–279, 73. Mitler MM, Seidel WF, van den Hoed J, et 42. Manabe K, Matsui T, Yamaya M, et al: 2003 al: Comparative hypnotic effects of flu- Sleep patterns and mortality among elderly razepam, triazolam, and placebo: a long- patients in a geriatric hospital. Gerontology 59. Kripke DF, Hauri P, Ancoli-Israel S, et al: term simultaneous nighttime and daytime 46:318–322, 2000 Sleep evaluation in chronic insomniacs dur- study. Journal of Clinical Psychopharmacol- ing 14-day use of flurazepam and midazo- 43. Leppavuori A, Pohjasvaara T, Vataja R, et ogy 4:2–13, 1984 lam. Journal of Clinical Psychopharmacolo- al: Insomnia in ischemic stroke patients. gy 10:32S–43S, 1990 74. IMS Health. National Prescription Audit Cerebrovascular Diseases 14:90–97, 2002 Plus. Fairfield, Conn, IMS Health, 2003 60. Kales A, Ansel RD, Markham CH, et al: 44. Mallon L, Broman JE, Hetta J: Sleep com- Sleep in patients with Parkinson’s disease 75. Mitler MM, Carskadon MA, Phillips RL, et plaints predict coronary artery disease mor- and normal patients prior to and following al: Hypnotic efficacy of temazepam: a long- tality in males: a 12-year follow-up study of levodopa administration. Clinical Pharma- term sleep laboratory evaluation. British a middle-aged Swedish population. Journal cology and Therapeutics 12:397–406, 1971 Journal of Clinical Pharmacology of Internal Medicine 251:207–216, 2002 8:63S–68S, 1979 45. Webb WB: Sleep in older persons: sleep 61. Vogel GW, Barker K, Gibbons P, et al: A structures of 50- to 60-year-old men and comparison of the effects of flurazepam 30 76. Roehrs T, Vogel G, Vogel F, et al: Dose ef- women. Journal of Gerontology 37:581– mg and triazolam 0.5 mg on the sleep of fects of temazepam tablets on sleep. Drugs 586, 1982 people with insomnia. Psychopharmacolo- Under Experimental and Clinical Research gy 47:81–86, 1976 12:693–699, 1986 46. Carskadon MA, Brown ED, Dement WC: Sleep fragmentation in the elderly: rela- 62. Elie R, Ruther E, Farr I, et al: Sleep laten- 77. Gilbert SS, Burgess HJ, Kennaway DJ, et tionship to daytime sleep tendency. Neuro- cy is shortened during 4 weeks of treatment al: Attenuation of sleep propensity, core hy- biology of Aging 3:321–327, 1982 with zaleplon, a novel nonbenzodiazepine pothermia, and peripheral heat loss after hypnotic. Journal of Clinical Psychiatry 60: temazepam tolerance. American Journal of 47. Morin CM, Hauri PJ, Espie CA, et al: Non- 536–544, 1999 Physiology. Regulatory, Integrative and pharmacologic treatment of chronic insom- Comparative Physiology 279:R1980– nia: an American Academy of Sleep Medi- 63. Scharf MB, Roth T, Vogel GW, et al: A mul- R1987, 2000 cine review. Sleep 22:1134–1156, 1999 ticenter, placebo-controlled study evaluat- ing zolpidem in the treatment of chronic in- 78. Ngen CC, Hassan R: A double-blind place- 48. McClusky HY, Milby JB, Switzer PK, et al: somnia. Journal of Clinical Psychiatry 55: bo-controlled trial of zopiclone 7.5 mg and Efficacy of behavioral versus triazolam 192–199, 1994 temazepam 20 mg in insomnia. Interna- treatment in persistent sleep-onset insom- tional Clinical Psychopharmacology 5:165– nia. American Journal of Psychiatry 148: 64. Fry J, Scharf M, Mangano R, et al: Zaleplon 171, 1990 121–126, 1991 improves sleep without producing rebound effects in outpatients with insomnia. Inter- 79. Mauri MC, Gianetti S, Pugnetti L, et al: 49. Hauri PJ: Can we mix behavioral therapy national Clinical Psychopharmacology 15: Quazepam versus triazolam in patients with with hypnotics when treating people with 141–152, 2000 sleep disorders: a double-blind study. Inter- insomnia? Sleep 20:1111–1118, 1997 national Journal of Clinical Pharmacology 65. Allen RP, Mendels J, Nevins DB, et al: Ef- Research 13:173–177, 1993 50. Morin CM, Colecchi C, Stone J, et al: Be- ficacy without tolerance or rebound insom- havioral and pharmacological therapies for nia for and temazepam after use 80. Ancoli-Israel S, Richardson GS, Mangano late-life insomnia: a randomized controlled for one to three months. Journal of Clinical RM: Long-term exposure to zaleplon is safe trial. JAMA 281:991–999, 1999 Pharmacology 27:768–775, 1987 and effective in younger-elderly and older-

342 PSYCHIATRIC SERVICES ♦ http://ps.psychiatryonline.org ♦ March 2005 Vol. 56 No. 3 elderly patients with primary insomnia. zodone, zolpidem, and triazolam in hu- 109.Witek TJ, Jr., Canestrari DA, Miller RD, Sleep 26:A77, 2003 mans. Psychopharmacology 144:220–233, et al: Characterization of daytime sleepi- 1999 ness and psychomotor performance fol- 81. Kummer J, Guendel L, Linden J, et al: lowing H1 receptor antagonists. Annals of Long-term polysomnographic study of the 95.Maxmen JS: Antidepressants, in Psy- Allergy, Asthma, and Immunology 74:419– efficacy and safety of zolpidem in elderly chotropic Drugs: Fast Facts 1st ed., New 426, 1995 psychiatric in-patients with insomnia. Jour- York, Norton, 1991 nal of International Medical Research 21: 110.Lessard E, Yessine MA, Hamelin BA, et al: 171–184, 1993 96.Janowsky D, Curtis G, Zisook S, et al: Ven- Diphenhydramine alters the disposition of tricular arrhythmias possibly aggravated venlafaxine through inhibition of CYP2D6 82. Walsh JK, Roth T, Randazzo A, et al: Eight by trazodone. American Journal of Psychi- activity in humans. Journal of Clinical Psy- weeks of non-nightly use of zolpidem for atry 140:796–797, 1983 chopharmacology 21:175–184, 2001 primary insomnia. Sleep 23:1087–1096, 2000 97.Thompson JW, Jr, Ware MR, Blashfield 111.Katzung BG: Basic and Clinical Pharma- RK: Psychotropic medication and pri- cology, 8th ed. New York, Lange Medical 83. Verster JC, Volkerts ER, Schreuder AH, et apism: a comprehensive review. Journal of Books/McGraw Hill, 2001 al: Residual effects of middle-of-the-night Clinical Psychiatry 51:430–433, 1990 administration of zaleplon and zolpidem on 112.Walsh JK, Muehlbach MJ, Lauter SA, et driving ability, memory functions, and psy- 98.Parrino L, Spaggiari MC, Boselli M, et al: al: Effects of triazolam on sleep, daytime chomotor performance. Journal of Clinical Clinical and polysomnographic effects of sleepiness, and morning stiffness in pa- Psychopharmacology 22:576–583, 2002 trazodone CR in chronic insomnia associ- tients with rheumatoid arthritis. Journal of ated with dysthymia. Psychopharmacology Rheumatology 23:245–252, 1996 84. Hesse LM, von Moltke LL, Greenblatt DJ: 116:389–395, 1994 Clinically important drug interactions with 113.Asnis GM, Chakraburtty A, DuBoff EA, et zopiclone, zolpidem and zaleplon. CNS 99.Scharf MB, Sachais BA: Sleep laboratory al: Zolpidem in SSRI-treated patients with Drugs 17:513–532, 2003 evaluation of the effects and efficacy of persistent insomnia. Journal of Clinical trazodone in depressed insomniac pa- Psychiatry 60:668–676, 1999 85. Ware JC, Walsh JK, Scharf MB, et al: Min- tients. Journal of Clinical Psychiatry imal rebound insomnia after treatment 51(suppl): 13–17, 1990 114.Haffmans PM, Vos MS: The effects of tra- with 10- mg zolpidem. Clinical Neurophar- zodone on sleep disturbances induced by macology 20:116–125, 1997 100.Saletu-Zyhlarz GM, Abu-Bakr MH, An- brofaromine. European Psychiatry 14: derer P, et al: Insomnia in depression: dif- 167–171, 1999 86. Saletu-Zyhlarz G, Anderer P, Brandstatter ferences in objective and subjective sleep N, et al: Placebo-controlled sleep laborato- and awakening quality to normal controls 115.Nierenberg AA, Adler LA, Peselow E, et ry studies on the acute effects of zolpidem and acute effects of trazodone. Progress in al: Trazodone for antidepressant-associat- on objective and subjective sleep and awak- Neuro-Psychopharmacology and Biologi- ed insomnia. American Journal of Psychia- ening quality in nonorganic insomnia relat- cal Psychiatry 26:249–260, 2002 try 151:1069–1072, 1994 ed to neurotic and stress-related disorder. 116.Nowlin-Finch NL, Altshuler LL, Szuba 101.Blacker R, Shanks NJ, Chapman N, et al: Neuropsychobiology 41:139–148, 2000 MP, et al: Rapid resolution of first episodes The drug treatment of depression in gen- of mania: sleep related? Journal of Clinical 87. Shaw SH, Curson H, Coquelin JP: A dou- eral practice: a comparison of nocte ad- Psychiatry 55:26–29, 1994 ble-blind, comparative study of zolpidem ministration of trazodone with , and placebo in the treatment of insomnia in dothiepin, and amitriptyline. Psychophar- 117.Morin CM, Stone J, McDonald K, et al: elderly psychiatric in-patients. Journal of macology 95(suppl):S18–S24, 1988 Psychological treatment of insomnia: a International Medical Research 20:150– clinical replication series with 100 pa- 102.Moon CA, Davey A: The efficacy and 161, 1992 tients. Behaviour Therapy 25:159–177, residual effects of trazodone (150 mg 1994 88. Wang PS, Bohn RL, Glynn RJ, et al: Zolpi- nocte) and mianserin in the treatment of dem use and hip fractures in older people. depressed general practice patients. Psy- 118.Spielman AJ, Saskin P, Thorpy MJ: Treat- Journal of the American Geriatrics Society chopharmacology 95(suppl):S7–13, 1988 ment of chronic insomnia by restriction of 49:1685–1690, 2001 time in bed. Sleep 10:45–56, 1987 103.Meuleman JR, Nelson RC, Clark RL, Jr.: 89. Mendelson WB: Clinical distinctions be- Evaluation of temazepam and diphenhy- 119.Espie CA, Lindsay WR, Brooks DN, et al: tween long-acting and short-acting benzo- dramine as hypnotics in a nursing-home A controlled comparative investigation of diazepines. Journal of Clinical Psychiatry population. Drug Intelligence and Clinical psychological treatments for chronic sleep 53(suppl):4–7, 1992 Pharmacy 21:716–720, 1987 onset insomnia. Behaviour Research and Therapy 27:79–88, 1989 90. Brassington GS, King AC, Bliwise DL: 104.2002 Sleep in America Poll. Washington, Sleep problems as a risk factor for falls in a DC, National Sleep Foundation, 2002 120.Leese P, Maier G: Eszopiclone: pharmaco- sample of community-dwelling adults aged kinetic (PK) and pharmacodynamic (PD) 64–99 years. Journal of the American Geri- 105.Rickels K, Morris RJ, Newman H, et al: effects of a novel anti-insomnia agent after atrics Society 48:1234–1240, 2000 Diphenhydramine in insomniac family daytime administration in healthy patients. practice patients: a double-blind study. Sleep 25:A45, 2002 91. Pies RW: Dose-related sensory distortions Journal of Clinical Pharmacology 23:234– with zolpidem. Journal of Clinical Psychia- 242, 1983 121.Krystal AD, Walsh JK, Laska E, et al: Sus- try 56:35–36, 1995 tained efficacy of eszopiclone over six 106.Kudo Y, Kurihara M: Clinical evaluation of months of nightly treatment: results of a 92. Hedner J, Yaeche R, Emilien G, et al: Zale- diphenhydramine hydrochloride for the randomized, double-blind, placebo con- plon shortens subjective sleep latency and treatment of insomnia in psychiatric pa- trolled study in adults with chronic insom- improves subjective sleep quality in elderly tients: a double-blind study. Journal of nia. Sleep 26:793–799, 2003 patients with insomnia. International Jour- Clinical Pharmacology 30:1041–1048, nal of Geriatric Psychiatry 15:704–712, 1990 122.Zammit G, Gillin JC, McNabb L, et al: Es- 2000 zopiclone, a novel nonbenzodiazepine 107.Richardson GS, Roehrs TA, Rosenthal L, anti-insomnia agent: a six-week efficacy 93. Haria M, Fitton A, McTavish D: Tra- et al: Tolerance to daytime sedative effects and safety study in adult patients with zodone: a review of its pharmacology, ther- of H1 antihistamines. Journal of Clinical chronic insomnia. Sleep 26:A297, 2003 apeutic use in depression, and therapeutic Psychopharmacology 22:511–515, 2002 potential in other disorders. Drugs and Ag- 123.Walsh JK, Lankford DD, Krystal A, et al: ing 4:331–355, 1994 108.Roth T, Roehrs T, Koshorek G, et al: Seda- Efficacy and tolerability of four doses of tive effects of antihistamines. Journal of indiplon (NBI-34060) modified-release in 94. Rush CR, Baker RW, Wright K: Acute be- Allergy and Clinical Immunology 80:94– elderly patients with sleep maintenance havioral effects and abuse potential of tra- 98, 1987 insomnia. Sleep 26:A78, 2003

PSYCHIATRIC SERVICES ♦ http://ps.psychiatryonline.org ♦ March 2005 Vol. 56 No. 3 343

Question 4

Evidence supporting the use of medical marijuana to treat or alleviate the disease or condition, including journal articles, peer-reviewed studies, and other types of medical or scientific documentation

Contents

Key Findings – 3

New Evidence –

Cannabidiol in Anxiety and Sleep: A Large Case Series – 5

Previously Submitted Evidence –

Effectiveness of Raw, Natural Medical Flower for Treating Insomnia under Naturalistic Conditions - 10

Patient-Reported Symptom Relief Following Consumption - 20

Cannabis, Pain, and Sleep: Lessons from Therapeutic Clinical Trials of Sativex, a Cannabis-Based Medicine - 29

Cannabidiol, a constituent of Cannabis sativa, modulates sleep in rats - 45

Endocannabinoid Signaling Regulates Sleep Stability – 55

Therapeutic Benefits of Cannabis: A Patient Survey - 103

Who Are Medical Marijuana Patients? Population Characteristics from Nine California Assessment Clinics - 108

Effectiveness of Cannabidiol Oil for Pediatric Anxiety and Insomnia as Part of Posttraumatic Stress Disorder: A Case Report - 118

Key Findings

Marijuana’s schedule I status makes studies on its medical use difficult to conduct. Because of this, evidence that medical marijuana is effective in treating insomnia is more limited than that of an FDA approved pharmaceutical, but significant evidence of medical marijuana’s effectiveness can still be seen. New Evidence

Cannabidiol in Anxiety and Sleep: A Large Case Series

Sleep scores improved within the first month in 48 patients (66.7%) but fluctuated over time. In this chart review, CBD was well tolerated in all but 3 patients.

Previously Submitted Evidence

Effectiveness of Raw, Natural Medical Cannabis Flower for Treating Insomnia under Naturalistic Conditions

Releaf app users experienced a statistically and clinically significant reduction in insomnia symptoms after using cannabis. Cannabis Indica strains and vaporizers in particular seemed the most effective ways of treating insomnia with cannabis.

Patient-Reported Symptom Relief Following Medical Cannabis Consumption

Cannabis users reporting their experiences to Releaf indicated a reduction of insomnia symptoms.

Cannabis, Pain, and Sleep: Lessons from Therapeutic Clinical Trials of Sativex, a Cannabis - Based Medicine

Sleep-laboratory results indicate a mild activating effect of CBD, and slight residual sedation with THC predominant extracts. Experience to date with Sativex in numerous Phase I – III studies in 2000 subjects with 1000 patient years of exposure demonstrate marked improvement in subjective sleep parameters in patients with a wide variety of pain conditions.

Cannabidiol, a constituent of Cannabis sativa, modulates sleep in rats

This study found that CBD modulates waking via activation of neurons in the hypothalamus and DRD. Both regions are apparently involved in the generation of alertness. Also, CBD increases DA levels as measured by microdialysis and HPLC procedures. Since CBD induces alertness, it might be of therapeutic value in sleep disorders such as excessive somnolence.

Endocannabinoid Signaling Regulates Sleep Stability

The major findings of this work are that eCB signaling through the CB1 receptor is necessary and sufficient for the stability of NREM sleep. Direct activation of CB1 with CP47 or increasing eCB tone with JZL or AM3506 augmented the time spent in NREM primarily due to increased NREM bout length. This suggests that increasing eCB signaling stabilizes the NREM state.

Therapeutic Benefits of Cannabis: A Patient Survey

45% of study participants reported relief from insomnia as a result of using cannabis as a treatment.

Who Are Medical Marijuana Patients? Population Characteristics from Nine California Assessment Clinics

70.7% of patients participating in this study reported therapeutic benefits which improved their sleep.

Effectiveness of Cannabidiol Oil for Pediatric Anxiety and Insomnia as Part of Posttraumatic Stress Disorder: A Case Report

The main finding from this case study is that CBD oil can be an effective compound to reduce anxiety and insomnia secondary to PTSD. A review of the literature suggests some benefits from the use of CBD because of its anxiolytic and sleep-inducing effects.9 Animal studies support use of this treatment and report that “CBD may block anxiety-induced [rapid eye movement] sleep alteration via its anxiolytic effect on the brain.”21

ORIGINAL RESEARCH & CONTRIBUTIONS Cannabidiol in Anxiety and Sleep: A Large Case Series Scott Shannon, MD1; Nicole Lewis, ND2; Heather Lee, PA-C3; Shannon Hughes, PhD4 Perm J 2019;23:18-041 E-pub: 01/07/2019 https://doi.org/10.7812/TPP/18-041

ABSTRACT and vomiting, and in Western medicine it was commonly used Context: Cannabidiol (CBD) is one of many cannabinoid com- as an analgesic.4,5 In the US, physicians prescribed Cannabis pounds found in cannabis. It does not appear to alter conscious- sativa for a multitude of illnesses until restrictions were put ness or trigger a “high.” A recent surge in scientific publications has in place in the 1930s and then finally stopped using it in 1970 found preclinical and clinical evidence documenting value for CBD when the federal government listed marijuana as a Schedule I in some neuropsychiatric disorders, including epilepsy, anxiety, substance, claiming it an illegal substance with no medical value. and schizophrenia. Evidence points toward a calming effect for California was the first state to go against the federal ban and CBD in the central nervous system. Interest in CBD as a treatment legalize medical marijuana in 1996.6 As of June 2018, 9 states of a wide range of disorders has exploded, yet few clinical studies and Washington, DC, have legalized recreational marijuana, of CBD exist in the psychiatric literature. and 30 states and Washington, DC, allow for use of medical Objective: To determine whether CBD helps improve sleep and/ marijuana.7 The purpose of the present study is to describe the or anxiety in a clinical population. effects of CBD on anxiety and sleep among patients in a clinic Design: A large retrospective case series at a psychiatric clinic presenting with anxiety or sleep as a primary concern. involving clinical application of CBD for anxiety and sleep com- CBD has demonstrated preliminary efficacy for a range of plaints as an adjunct to usual treatment. The retrospective chart physical and mental health care problems. In the decade before review included monthly documentation of anxiety and sleep 2012, there were only 9 published studies on the use of canna- quality in 103 adult patients. binoids for medicinal treatment of pain; since then, 30 articles Main Outcome Measures: Sleep and anxiety scores, using have been published on this topic, according to a PubMed search validated instruments, at baseline and after CBD treatment. conducted in December 2017. Most notable was a study con- Results: The final sample consisted of 72 adults presenting ducted at the University of California, San Diego’s Center for with primary concerns of anxiety (n = 47) or poor sleep (n = 25). Medicinal Cannabis Research that showed cannabis cigarettes Anxiety scores decreased within the first month in 57 patients reduced pain by 34% to 40% compared with placebo (17% to (79.2%) and remained decreased during the study duration. Sleep 20% decrease in pain).8 In particular, CBD appears to hold scores improved within the first month in 48 patients (66.7%) but benefits for a wide range of neurologic disorders, including fluctuated over time. In this chart review, CBD was well tolerated decreasing major seizures. A recent large, well-controlled study in all but 3 patients. of pediatric epilepsy documented a beneficial effect of CBD in Conclusion: Cannabidiol may hold benefit for anxiety-related reducing seizure frequency by more than 50%.9 In addition to disorders. Controlled clinical studies are needed. endorphin release, the “runner’s high” experience after exercise has been shown to be induced in part by anandamide acting on INTRODUCTION CB1 receptors, eliciting anxiolytic effects on the body.10 The ac- TheCannabis plant has been cultivated and used for its me- tivity of CBD at 5-HT1A receptors may drive its neuroprotective, dicinal and industrial benefits dating back to ancient times. antidepressive, and anxiolytic benefits, although the mechanism Cannabis sativa and Cannabis indica are the 2 main species.1 of action by which CBD decreases anxiety is still unclear.11 The Cannabis plant contains more than 80 different chemicals CBD was shown to be helpful for decreasing anxiety through known as cannabinoids. The most abundant cannabinoid, tet- a simulated public speaking test at doses of 300 mg to 600 mg rahydrocannabinol (THC), is well known for its psychoactive in single-dose studies.12-14 Other studies suggest lower doses of properties, whereas cannabidiol (CBD) is the second-most 10 mg/kg having a more anxiolytic effect than higher doses of abundant and is nonpsychoactive. Different strains of the 100 mg/kg in rats.15 A crossover study comparing CBD with plant are grown containing varying amounts of THC and found that high-dose CBD at 160 mg increased CBD. Hemp plants are grown for their fibers and high levels the duration of sleep.16 Another crossover study showed that of CBD that can be extracted to make oil, but marijuana plants grown for recreational use have higher concentrations of THC compared with CBD.2 Industrial hemp must contain less than 0.3% THC to be considered legal, and it is from this plant that 3 Author Affiliations CBD oil is extracted. 1 Department of Psychiatry, University of Colorado, Denver Many different cultures have used theCannabis plant to treat 2 Department of Naturopathic Medicine, Wholeness Center, Fort Collins, CO a plethora of ailments. Practitioners in ancient China targeted 3 North Range Behavioral Health, Greeley, CO malaria, menstrual symptoms, gout, and constipation. During 4 School of Social Work, Colorado State University College of Health and Human Sciences, Fort Collins medieval times, cannabis was used for pain, epilepsy, nausea, Corresponding Author Scott Shannon, MD ([email protected])

Keywords: anxiety, cannabidiol, CBD, sleep

The Permanente Journal • https://doi.org/10.7812/TPP/18-041 1 ORIGINAL RESEARCH & CONTRIBUTIONS Cannabidiol in Anxiety and Sleep: A Large Case Series

plasma cortisol levels decreased more significantly when given and schizoaffective disorder received a CBD dosage that was oral CBD, 300 to 600 mg, but these patients experienced a seda- gradually increased to 175 mg/d. tive effect.17 The higher doses of CBD that studies suggest are Often CBD was employed as a method to avoid or to re- therapeutic for anxiety, insomnia, and epilepsy may also increase duce psychiatric medications. The CBD selection and dosing mental sedation.16 Administration of CBD via different routes reflected the individual practitioner’s clinical preference. In- and long-term use of 10 mg/d to 400 mg/d did not create a formed consent was obtained for each patient who was treated toxic effect on patients. Doses up to 1500 mg/d have been well and considered for this study. Monthly visits included clinical tolerated in the literature.18 Most of the research done has been evaluation and documentation of patients’ anxiety and sleep sta- in animal models and has shown potential benefit, but clinical tus using validated measures. CBD was added to care, dropped data from randomized controlled experiments remain limited. from care, or refused as per individual patient and practitioner Finally, the most notable benefit of cannabis as a form of preference. The Western Institutional Review Board, Puyallup, treatment is safety. There have been no reports of lethal overdose WA, approved this retrospective chart review. with either of the cannabinoids and, outside of concerns over abuse, major complications are very limited.19 Current research Setting and Sample indicates that cannabis has a low overall risk with short-term Wholeness Center is a large mental health clinic in Fort Col- use, but more research is needed to clarify possible long-term lins, CO, that focuses on integrative medicine and psychiatry. risks and harms. Practitioners from a range of disciplines (psychiatry, naturopa- Given the promising biochemical, physiologic, and preclinical thy, acupuncture, neurofeedback, yoga, etc) work together in data on CBD, a remarkable lack of randomized clinical trials a collaborative and cross-disciplinary environment. CBD had and other formal clinical studies exist in the psychiatric arena. been widely incorporated into clinical care at Wholeness Center The present study describes a series of patients using CBD for a few years before this study, on the basis of existing research treatment of anxiety or sleep disturbances in a clinical practice and patient experience. setting. Given the paucity of data in this area, clinical observa- The sampling frame consisted of 103 adult patients who were tions can be quite useful to advance the knowledge base and consecutively treated with CBD at our psychiatric outpatient to offer questions for further investigation. This study aimed to clinic. Eighty-two (79.6%) of the 103 adult patients had a determine whether CBD is helpful for improving sleep and/ documented anxiety or sleep disorder diagnosis. Patients with or anxiety in a clinical population. Given the novel nature of sole or primary diagnoses of schizophrenia, posttraumatic stress this treatment, our study also focused on tolerability and safety disorder, and agitated depression were excluded. Ten patients concerns. As a part of the evolving legal status of cannabis, our were further excluded because they had only 1 documented investigation also looked at patient acceptance. visit, with no follow-up assessment. The final sample consisted of 72 adult patients presenting with primary concerns of anxiety METHODS (65.3%; n = 47) or poor sleep (34.7%; n = 25) and who had at Design and Procedures least 1 follow-up visit after CBD was prescribed. A retrospective chart review was conducted of adult psychiat- ric patients treated with CBD for anxiety or sleep as an adjunct Main Outcome Measures to treatment as usual at a large psychiatric outpatient clinic. Sleep and anxiety were the targets of this descriptive re- Any current psychiatric patient with a diagnosis by a mental port. Sleep concerns were tracked at monthly visits using the health professional (psychiatrist, psychiatric nurse practitioner, Pittsburg Sleep Quality Index. Anxiety levels were monitored or physician assistant) of a sleep or anxiety disorder was con- at monthly visits using the Hamilton Anxiety Rating Scale. sidered. Diagnosis was made by clinical evaluation followed by Both scales are nonproprietary. The Hamilton Anxiety Rating baseline psychologic measures. These measures were repeated Scale is a widely used and validated anxiety measure with 14 monthly. Comorbid psychiatric illnesses were not a basis for individual questions. It was first used in 1959 and covers a wide exclusion. Accordingly, other psychiatric medications were range of anxiety-related concerns. The score ranges from 0 to 56. administered as per routine patient care. Selection for the case A score under 17 indicates mild anxiety, and a score above 25 series was contingent on informed consent to be treated with indicates severe anxiety. The Pittsburg Sleep Quality Index is a CBD for 1 of these 2 disorders and at least 1 month of active self-report measure that assesses the quality of sleep during a treatment with CBD. Patients treated with CBD were provided 1-month period. It consists of 19 items that have been found to with psychiatric care and medications as usual. Most patients be reliable and valid in the assessment of a range of sleep-related continued to receive their psychiatric medications. The patient problems. Each item is rated 0 to 3 and yields a total score from population mirrored the clinic population at large with the 0 to 21. A higher number indicates more sleep-related concerns. exception that it was younger. A score of 5 or greater indicates a “poor sleeper.” Nearly all patients were given CBD 25 mg/d in capsule form. Side effects and tolerability of CBD treatment were assessed If anxiety complaints predominated, the dosing was every morn- through spontaneous patient self-reports and were documented ing, after breakfast. If sleep complaints predominated, the dosing in case records. Any other spontaneous comments or complaints was every evening, after dinner. A handful of patients were given of patients were also documented in case records and included CBD 50 mg/d or 75 mg/d. One patient with a trauma history in this analysis.

2 The Permanente Journal • https://doi.org/10.7812/TPP/18-041 ORIGINAL RESEARCH & CONTRIBUTIONS Cannabidiol in Anxiety and Sleep: A Large Case Series

Data Analysis Deidentified patient data were evaluated using descriptive statistics and plotted graphically for visual analysis and inter- pretation of trends. RESULTS The average age for patients with anxiety was 34 years (range = 18-70 years) and age 36.5 years for patients with sleep disorders (range = 18-72 years). Most patients with an anxiety di- agnosis were men (59.6%, 28/47), whereas more sleep-disordered patients were women (64.0%, 16/25). All 72 patients completed sleep and anxiety assessments at the onset of CBD treatment and at the first monthly follow-up. By the second monthly follow-up, 41 patients (56.9%) remained on CBD treatment and completed Figure 1. Mean anxiety and sleep scores for adults using cannabidiol treatment. assessments; 27 patients (37.5%) remained on CBD treatment at the third monthly assessment. HAM-A = Hamilton Anxiety Rating Scale; PSQI = Pittsburg Sleep Quality Index. Table 1 provides means and standard deviations for sleep and anxiety scores at baseline and during the follow-up period for CBD was well tolerated, with few patients reporting side ef- adults taking CBD. Figure 1 graphically displays the trend in fects. Two patients discontinued treatment within the first week anxiety and sleep scores over the study period. On average, anxiety because of fatigue. Three patients noted mild sedation initially that and sleep improved for most patients, and these improvements appeared to abate in the first few weeks. One patient with a devel- were sustained over time. At the first monthly assessment after opmental disorder (aged 21 years) had to be taken off the CBD the start of CBD treatment, 79.2% (57/72) and 66.7% (48/72) regimen because of increased sexually inappropriate behavior. of all patients experienced an improvement in anxiety and sleep, The CBD was held, and the behavior disappeared. The behavior respectively; 15.3% (11/72) and 25.0% (18/72) experienced reappeared on redosing 2 weeks later, and the CBD regimen was worsening symptoms in anxiety and sleep, respectively. Two formally discontinued. The treating psychiatrist thought this was months after the start of CBD treatment, 78.1% (32/41) and related to disinhibition because the patient’s anxiety responded 56.1% (23/41) of patients reported improvement in anxiety and dramatically. One patient noted dry eyes. Reasons for patients sleep, respectively, compared with the prior monthly visit; again, not following-up at later assessment points are largely unknown 19.5% (8/41) and 26.8% (11/41), respectively, reported worsening but are probably because of standard attrition experienced in problems as compared with the prior month. usual clinical practice. There was no evidence to suggest patients These results demonstrated a more sustained response to anxi- discontinued care because of tolerability concerns. The attrition ety than for sleep over time. Patient records displayed a larger rates were similar in nature and size to those found in routinely decrease in anxiety scores than in sleep scores. The sleep scores scheduled visits in this clinic. demonstrated mild improvement. The anxiety scores decreased The treatment with CBD was in general well accepted, as within the first month and then remained decreased during the judged by the clinicians’ and patients’ responses. Four patients study duration. declined CBD treatment because of religious or ethical con- cerns about the relation to cannabis. Nearly all patients easily provided informed consent once the nature of the treatment Table 1. Descriptive statistics for anxiety and sleep scores was explained. Most patients appreciated the opportunity to among adults using cannabidiol treatment try something natural and avoid further or initial psychiatric Parameter HAM-A, mean (SD) PSQI, mean (SD) medication use. Anxiety (n = 47) DISCUSSION Baseline 23.87 (9.87) 10.98 (3.43) In an outpatient psychiatric population, sleep scores displayed 1-month follow-up 18.02 (7.56) 8.88 (3.68) no sustained improvements during the 3-month study. Anxiety 2-month follow-up 16.35 (8.80) 8.59 (2.91) scores decreased fairly rapidly, and this decrease was sustained 3-month follow-up 16.36 (9.80) 9.25 (2.46) during the study period. These results are consistent with the Sleep disorder (n = 25) existing preclinical and clinical data on CBD. CBD was well Baseline 22.18 (7.55) 13.08 (3.03) accepted and well tolerated in our patients. Side effects were 1-month follow-up 17.82 (9.72) 10.64 (3.89) minimal (mainly fatigue) and may be related to dosing. 2-month follow-up 17.36 (10.91) 9.39 (3.81) The doses used in this study (25 mg/d to 175 mg/d) were 3-month follow-up 13.78 (7.86) 9.33 (4.63) much lower than those reported in some of the clinical literature 12-14,17 HAM-A = Hamilton Anxiety Rating Scale; PSQI = Pittsburg Sleep Quality Index; (300 mg/d to 600 mg/d) for 2 reasons. The first is that in SD = standard deviation. our experience lower doses appear to elicit an adequate clinical

The Permanente Journal • https://doi.org/10.7812/TPP/18-041 3 ORIGINAL RESEARCH & CONTRIBUTIONS Cannabidiol in Anxiety and Sleep: A Large Case Series

response. Second, the current retail cost of CBD would make understanding of the physiology and neurologic pathways the use of 600 mg/d cost prohibitive. points to a benefit with anxiety-related issues. The results of our clinical report support the existing scientific evidence. In Study Limitations our study, we saw no evidence of a safety issue that would limit These results must be interpreted cautiously because this future studies. In this evaluation, CBD appears to be better was a naturalistic study, all patients were receiving open-label tolerated than routine psychiatric medications. Furthermore, treatment, and there was no comparison group. Concurrent psy- CBD displays promise as a tool for reducing anxiety in clini- chiatric medications were employed as in routine clinical care. cal populations, but given the open-label and nonrandomized This is both a limitation and strength, as very few publications nature of this large case series, all results must be interpreted exist in this population. Other researchers have noted that the very cautiously. Randomized and controlled trials are needed v large societal notoriety about cannabis and medical marijuana to provide definitive clinical guidance. probably contributes to a larger-than-normal placebo effect.20 Any study that explores efficacy in this therapy probably will Disclosure Statement Dr Shannon has published several professional books on integrative struggle with a potentially inflated placebo effect that will mental health. Dr Shannon is a Principal Investigator for a Phase 3 study of make these determinations more difficult. Likewise, the clinical 3,4-methylenedioxy-methamphetamine (MDMA)-assisted psychotherapy for population in this case series is skewed younger than typical for severe posttraumatic stress disorder and receives compensation for his clinical our clinic, and future studies could explore the possible selec- work from the Multidisciplinary Association for Psychedelic Studies, Santa Cruz, tion bias inherent in this treatment option. Most patients were CA. The other authors have no conflicts of interests to disclose. also taking psychiatric medications and receiving other mental health services, such as counseling, which limits the ability to Acknowledgments make any causal links to CBD treatment. Clinical attrition is CV Sciences Inc, Las Vegas, NV, provided cannabidiol products for the evident in the dataset. The reason for this might be related to study. CV Sciences was not involved in the data collection, data interpretation, preparation of the report, or decision to submit the report for publication. No CBD ingestion or not, so the overall component remains un- other financial support was provided. The authors would like to express their clear. Furthermore, patients at our clinic often express a desire deep appreciation to the staff and clinicians at Wholeness Center for their to reduce or to avoid use of psychiatric medications, which professionalism. may contribute to an enhanced placebo effect or additional Kathleen Louden, ELS, of Louden Health Communications provided editorial bias. The length of clinical monitoring may help to decrease assistance. this concern. However, the clinical data in this analysis show How to Cite this Article a trend toward clinically significant relief of anxiety upon the Shannon S, Lewis N, Lee H, Hughes S. Cannabidiol in anxiety and sleep: start of CBD treatment. A large case series. Perm J 2019;23:18-041. DOI: https://doi.org/10.7812/ Legality of Cannabidiol TPP/18-041 The legality of CBD is not clear. Like the issues surrounding References the legality of cannabis in general, CBD presents the clinician 1. Baron EP. Comprehensive review of medicinal marijuana, cannabinoids, and with a confusing state vs federal legal quandary, and this keeps therapeutic implications in medicine and headache: What a long strange trip the issue in question. CBD is legal in the 33 states that have it’s been …. Headache 2015 Jun;55(6):885-916. DOI: https://doi.org/10.1111/ head.12570. legalized medical or recreational use of marijuana and in 17 2. Schluttenhofer C, Yuan L. Challenges towards revitalizing hemp: A multifaceted other states that have legalized some form of CBD, according crop. Trends Plant Sci 2017 Nov;22(11):917-29. DOI: https://doi.org/10.1016/j. to the National Organization for the Reform of Marijuana tplants.2017.08.004. 21 3. Andre CM, Hausman JF, Guerriero G. Cannabis sativa: The plant of the thousand Laws (NORML). But like marijuana, it is still not legal at and one molecules. Front Plant Sci 2016 Feb 4;7:19. DOI: https://doi.org/10.3389/ the federal level. The federal government has announced that fpls.2016.00019. it is not focused on this compound in terms of enforcement 4. Zlebnik NE, Cheer JF. Beyond the CB1 receptor: Is cannabidiol the answer for 22 disorders of motivation? Annu Rev Neurosci 2016 Jul 8;39:1-17. DOI: https://doi. or interdiction. However, CBD is interpreted by the Drug org/10.1146/annurev-neuro-070815-014038. Enforcement Administration, Food and Drug Administration, 5. Devinsky O, Cilio MR, Cross H, et al. Cannabidiol: Pharmacology and potential and Congress to be a Schedule I substance, and therefore it is therapeutic role in epilepsy and other neuropsychiatric disorders. Epilepsia 2014 illegal in all 50 states.23 Pragmatically, CBD is widely avail- Jun;55(6):791-802. DOI: https://doi.org/10.1111/epi.12631. 6. Bostwick JM. Blurred boundaries: The therapeutics and politics of medical able on the Internet, with sales expected to reach $1 billion by marijuana. Mayo Clin Proc 2012 Feb;87(2):172-86. DOI: https://doi.org/10.1016/j. 2020. Pending federal legislation to redefine the legal status of mayocp.2011.10.003. cannabis would clarify this complex issue. Canada’s move to 7. Legal recreational marijuana states and DC: Cannabis laws with possession and cultivation limits [Internet]. Santa Monica, CA: ProCon.org; 2018 Jun 27 legalize cannabis in October 2018 further highlights the need [cited 2018 Aug 23]. Available from: https://marijuana.procon.org/view.resource. for a speedy resolution to this question.24 php?resourceID=006868. 8. Abrams DI, Jay CA, Shade SB, et al. Cannabis in painful HIV-associated sensory CONCLUSION neuropathy: A randomized placebo-controlled trial. Neurology 2007 Feb 13;68(7):515- 21. DOI: https://doi.org/10.1212/01.wnl.0000253187.66183.9c. Formal studies on efficacy and dose finding are much needed. 9. Devinsky O, Cross JH, Laux L, et al; Cannabidiol in Dravet Syndrome Study Group. Some urgency exists, given the explosion of lay interest in Trial of cannabidiol for drug-resistant seizures in the Dravet syndrome. N Engl J Med 2017 May 25;376(21):2011-20. DOI: https://doi.org/10.1056/NEJMoa1611618. this topic and the rush to market these compounds. Current

4 The Permanente Journal • https://doi.org/10.7812/TPP/18-041 ORIGINAL RESEARCH & CONTRIBUTIONS Cannabidiol in Anxiety and Sleep: A Large Case Series

10. Fuss J, Steinle J, Bindila L, et al. A runner’s high depends on cannabinoid receptors 17. Zuardi AW, Guimarães FS, Moreira AC. Effect of cannabidiol on plasma prolactin, in mice. Proc Natl Acad Sci U S A 2015 Oct 20;112(42):13105-8. DOI: https://doi. growth hormone and cortisol in human volunteers. Braz J Med Biol Res 1993 org/10.1073/pnas.1514996112. Feb;26(2):213-7. 11. Zanelati TV, Biojone C, Moreira FA, Guimarães FS, Joca SR. Antidepressant- 18. Iffland K, Grotenhermen F. An update on safety and side effects of cannabidiol: A like effects of cannabidiol in mice: Possible involvement of 5-HT1A receptors. Br review of clinical data and relevant animal studies. Cannabis Cannabinoid Res 2017 J Pharmacol 2010 Jan;159(1):122-8. DOI: https://doi.org/10.1111/j.1476- Jun 1;2(1):139-54. DOI: https://doi.org/10.1089/can.2016.0034. 5381.2009.00521.x. 19. Collen M. Prescribing cannabis for harm reduction. Harm Reduct J 2012 Jan 1;9:1. 12. Zuardi AW, Rodrigues NP, Silva AL, et al. Inverted U-shaped dose-response curve of DOI: https://doi.org/10.1186/1477-7517-9-1. the anxiolytic effect of cannabidiol during public speaking in real life. Front Pharmacol 20. Loflin MJE, Earleywine M, Farmer S, Slavin M, Luba R, Bonn-Miller M. Placebo 2017 May 11;8:259. DOI: https://doi.org/10.3389/fphar.2017.00259. effects of edible cannabis: Reported intoxication effects at a 30-minute delay. J 13. Bergamaschi MM, Queiroz RH, Chagas MH, et al. Cannabidiol reduces the anxiety Psychoactive Drugs 2017 Nov-Dec;49(5):393-7. DOI: https://doi.org/10.1080/027910 induced by simulated public speaking in treatment-naïve social phobia patients. 72.2017.1354409. Neuropsychopharmacology 2011 May;36(6):1219-26. DOI: https://doi.org/10.1038/ 21. State laws [Internet]. Washington, DC: NORML Foundation; c2018 [cited 2018 Aug 7]. npp.2011.6. Available from: https://norml.org/laws. 14. Zuardi AW, Cosme RA, Graeff FG, Guimarães FS. Effects of ipsapirone 22. Mitchell T. Did the DEA just quietly approve CBD? [Internet]. 2018 Jun 2 [cited 2018 andcannabidiol on human experimental anxiety. J Psychopharmacol 1993 Jan;7(1 Aug 7]. Denver, CO: Westword; Available from: www.westword.com/marijuana/dea- Suppl):82-8. DOI: https://doi.org/10.1177/026988119300700112. quietly-gives-cbd-other-non-psychoactive-cannabinoids-the-go-ahead-10377016. 15. Guimarães FS, Chiaretti TM, Graeff FG, Zuardi AW. Antianxiety effect of cannabidiol 23. Clarification of the new drug code (7350) for marijuana extract [Internet]. Springfield, in the elevated plus-maze. Psychopharmacology (Berl) 1990;100(4):558-9. https:// VA: Drug Enforcement Administration, Diversion Control Division; 2017 [cited 2018 doi.org/10.1007/bf02244012. Aug 7]. Available from: www.deadiversion.usdoj.gov/schedules/marijuana/m_ 16. Zhornitsky S, Potvin S. Cannabidiol in humans—the quest for therapeutic extract_7350.html. targets. Pharmaceuticals (Basel) 2012 May 21;5(5):529-52. DOI: https://doi. 24. Steinmetz K. What marijuana legalization in Canada could mean for the United States org/10.3390/ph5050529. [Internet]. New York, NY: Time; 2017 Apr 6 [cited 2018 Aug 7]. Available from: http:// time.com/4728091/canada-legalizing-marijuana-united-states-weed-pot/.

The Permanente Journal • https://doi.org/10.7812/TPP/18-041 5 medicines

Article Effectiveness of Raw, Natural Medical Cannabis Flower for Treating Insomnia under Naturalistic Conditions

Jacob M. Vigil 1,*, Sarah S. Stith 2, Jegason P. Diviant 1, Franco Brockelman 3, Keenan Keeling 3 and Branden Hall 3 1 Department of Psychology, University of New Mexico, Albuquerque, NM 87131, USA; [email protected] 2 Department of Economics, University of New Mexico, Albuquerque, NM 87131, USA; [email protected] 3 Morebetter Ltd., Washington, DC 20012, USA; [email protected] (F.B.); [email protected] (K.K.); [email protected] (B.H.) * Correspondence: [email protected]; Tel.: +1-505-277-0374

 Received: 1 June 2018; Accepted: 9 July 2018; Published: 11 July 2018 

Abstract: Background: We use a mobile software application (app) to measure for the first time, which fundamental characteristics of raw, natural medical Cannabis flower are associated with changes in perceived insomnia under naturalistic conditions. Methods: Four hundred and nine people with a specified condition of insomnia completed 1056 medical cannabis administration sessions using the Releaf AppTM educational software during which they recorded real-time ratings of self-perceived insomnia severity levels prior to and following consumption, experienced side effects, and product characteristics, including combustion method, cannabis subtypes, and/or major cannabinoid contents of cannabis consumed. Within-user effects of different flower characteristics were modeled using a fixed effects panel regression approach with standard errors clustered at the user level. Results: Releaf AppTM users showed an average symptom severity reduction of −4.5 points on a 0–10 point visual analogue scale (SD = 2.7, d = 2.10, p < 0.001). Use of pipes and vaporizers was associated with greater symptom relief and more positive and context-specific side effects as compared to the use of joints, while vaporization was also associated with lower negative effects. Cannabidiol (CBD) was associated with greater statistically significant symptom relief than tetrahydrocannabinol (THC), but the cannabinoid levels generally were not associated with differential side effects. Flower from C. sativa plants was associated with more negative side effects than flower from C. indica or hybrid plant subtypes. Conclusions: Consumption of medical Cannabis flower is associated with significant improvements in perceived insomnia with differential effectiveness and side effect profiles, depending on the product characteristics.

Keywords: insomnia; Cannabis; marijuana; sleep; sleep disturbance; flower; cannabidiol; tetrahydrocannabinol; C. indica; C. sativa

1. Introduction Nearly 50% of the adult population in the United States (US) experiences sleeping problems [1–4], with high rates of dissatisfaction over the effectiveness and potential side effects of conventional pharmaceutical sleep aid medications [5]. Prescription sleep aids, including antidepressants, benzodiazepines, gamma-aminobutyric acid (GABA) medications, and anti-psychotics, are associated with significant negative side effects and risks of dangerous drug interactions [6]. Over-the-counter (OTC) medications such as antihistamines, melatonin, and , are generally less dangerous than prescription pharmaceuticals, but also less effective, and can also carry negative consequences (e.g., headaches, confusion, agitation), including residual effects (e.g., drowsiness, difficulty

Medicines 2018, 5, 75; doi:10.3390/medicines5030075 www.mdpi.com/journal/medicines Medicines 2018, 5, 75 2 of 10 concentrating, and memory impairments) that can lead to secondary problem behaviors (e.g., lethargy, work absenteeism) [2,7–11]. First generation antihistamines (e.g., diphenhydramine, hydroxyzine, succinate, and clemastine) act as anticholinergics. They generally score high on the anticholinergic cognitive burden scale. Alone or coupled with other anticholinergics, these medications, in addition to other classes of muscarinic antagonists, may contribute to an increased risk of developing dementia because the effects are cumulative and the body’s production of acetylcholine diminishes with age [12]. These circumstances may make people with insomnia willing to experiment with alternative sleep aid therapies, including medical cannabis, which is commonly used for treating insomnia [13–15] and becoming increasingly accessible due to expanding medical and now recreational cannabis reformation laws. Clinical randomized controlled trials (RCTs) have historically used cannabis extracts and synthetic phytocannabinoid analogues for estimating the pharmacodynamic properties of cannabis used in vivo [16]. They generally show that while cannabis consumption can improve some sleep outcomes (e.g., latency to sleep, reduced REM sleep problem behaviors), it is often associated with negative (albeit, relatively minor) side effects, and the major phytocannabinoids, delta-9 tetrahydrocannabinol (THC) and cannabidiol (CBD), may have different effects at different stages of sleep [17–22]. The improved sleep outcomes associated with THC may be influenced by the putative interaction of the CB1 cannabinoid receptor with orexin receptors that are, in turn, targeted by sleep aid medications such as [23]. Unfortunately, RCTs are poorly suited for assessing real-life patient decisions and capturing the phasic and inconsistent nature of the most common type of cannabis products used by millions of people daily, raw whole natural dried Cannabis flowers [16,24]. No study to date has measured the relative associations between THC and CBD contents and other basic characteristics of natural Cannabis flower (e.g., route of administration, cannabis subtypes) consumption and self-perceived feelings of insomnia in real-time under naturalistic conditions. This is the first study to measure which fundamental attributes of commonly consumed dried Cannabis flower affect perceived insomnia levels and experienced side effects. We operationalize our research question using a mobile educational software application (app) [25] for recording how combustion method, cannabis subtypes, and major cannabinoid contents are associated with real-time measurements of subjective insomnia levels, prior to and following administration of cannabis, and the manifestation of myriad possible side effects from normative use in users’ natural environments.

2. Materials and Methods

2.1. Study Design Institutional Review Board exemption was obtained from the University of New Mexico for this study, and the data were obtained from MoreBetter Ltd. (Washington, DC, USA), subject to a confidentiality agreement. This study uses self-collected data and user experiences recorded with the Releaf AppTM (version 1.4.1, Morebetter LTD, Washington, DC, USA) between June 2016 and May 2018. This mobile device educational software application was designed to track the effects of different types of medical cannabis products used under normal circumstances in natural environments, so that users can more optimally treat their underlying medical condition, with insomnia as one possibly treatable condition. Users of this version of the app consented through a privacy agreement for their anonymous data to be statistically analyzed and published in aggregate form. Prior to beginning each session, the user specifies the condition to be treated and the starting symptom level as well as a broad range of product characteristics. (The user interface for the Releaf AppTM is shown in Supplemental Figure S1). For the purpose of uniformity, we include only users of Cannabis flower. For analyses including THC and CBD, we omitted observations with labels indicated THC potencies greater than 35% or CBD potencies greater than 30%, because higher percentages do not occur naturally in flower. Medicines 2018, 5, 75 3 of 10

2.2. Study Outcomes Our study outcomes focus on symptom relief (reductions in insomnia symptom severity) and side effects. Product characteristic entry is voluntary, so the number of observations varies depending on the product characteristics included. Once a session has begun, the user can update their symptom level at any time. Symptom levels range from 0 (no detectable level) to 10 (most severe intensity level). In our analysis, we include only users reporting an initial symptom level of 1 or higher. Symptom Relief is measured as the starting symptom level minus the ending symptom level and ranges between −10 (maximum symptom relief) and 9 (minimum possible symptom relief). Our other outcome variables are any side effect reported by category (negative, positive, and context-specific) and percent of total available side effects in that category with 13 negative side effects, 19 positive side effects, and 10 context-specific side effects available for selection at any time during the session. (The side effects with categories are listed in Supplemental Table S1).

2.3. Statistical Analysis We use ordinary least squares panel regressions to analyze the effects of product characteristics on symptom relief and side effect profiles. We regress symptom relief on the product characteristics separately. The sample size changes due to non-reporting of product characteristics. For our six separate side effect outcomes, we include all the product characteristics together for the sake of brevity. Because starting symptom levels are a strong predictor of symptom relief, partly mechanistically (e.g., higher starting symptoms enable greater possible symptom relief), we include the starting symptom level in all our regressions. Although the outcomes are [0, 1] in the “any” side effect regressions and [0, 1] in the “percent” side effect regressions, we use ordinary least squares for the sake of consistency across regression models with different outcomes. All regressions include time-invariant user fixed effects and standard errors are clustered at the user level to account for heteroskedasticity and arbitrary correlation at the user level. Analyses were conducted using Stata 13.1.

3. Results Four hundred and nine users reported using flower to treat insomnia during 1056 sessions. In addition to the type of product, the user is prompted to report the combustion method (joint [13%], pipe [38%], and vape [49%]), plant subtypes (C. Indica [60%], C. Sativa [6%], and hybrid [33%]), and THC and CBD content (percentage of total weight). The mean THC level was 20% (SD = 5.39 percentage points) and the mean CBD level was 5.7% (SD = 5.44 percentage points). Starting symptom levels average 6.6 (SD = 2.1), while ending symptom levels average 2.2 (SD = 2.1). Mean symptom relief was −4.5 (standard deviation = 2.7, d = 2.10, p < 0.001). During the average session, users in our sample report 10% of negative side effects, 21% of positive side effects, and 24% of context-specific side effects. In addition, in 57% of sessions, users report at least one negative side effect, in 95% at least one positive side effect, and in 86% at least one context specific side effect. Table1 presents complete descriptive statistics. Additionally, insomnia patients reported using 461 different strains. Among the most frequently used strains, there was wide variability in cannabinoid contents, with two C. indica strains with fairly high THC (around 20%) and high (20%) to moderate (7%) CBD potencies (“Granddaddy Purple” and “Northern Lights”), followed by two hybrid strains with similar amounts of THC but CBD potencies of less than 4% (“OG Kush” and “Blue Dream”). Medicines 2018, 5, 75 4 of 10

Table 1. Descriptive Statistics.

Variable Mean Std. Dev Minimum Maximum Panel A: Subtypes (983 sessions, 378 users) Hybrid 0.33 0.47 0 1 C. indica 0.60 0.49 0 1 C. sativa 0.06 0.24 0 1 Panel B: Combustion Method (996 sessions, 385 users) Joint 0.13 0.34 0 1 Pipe 0.38 0.48 0 1 Vape 0.49 0.50 0 1 Panel C: THC (353 sessions, 143 users) % THC 0.19 0.54 0.02 0.35 THC < 10% 0.05 0.22 0 1 THC 10–19% 0.50 0.50 0 1 THC 20–34% 0.45 0.50 0 1 Panel D: CBD (119 sessions, 281 users) % CBD 0.60 0.54 0 0.30 CBD 0% 0.19 0.39 0 1 CBD 1–9% 0.51 0.50 0 1 CBD 10–34% 0.30 0.46 0 1 Panel E: Outcome and Control Variables (1056 sessions, 409 users) Symptom Change −4.5 2.7 −10 9 Starting Symptom Level 6.6 2.1 1 10 Ending Symptom Level 2.2 2.1 0 10 Panel F: Side Effects (1215 sessions, 359 users) Any Negative Side Effect 0.57 0.50 0 1 % of Negative Side Effects 0.10 0.13 0 1 Any Positive Side Effect 0.95 0.23 0 1 % of Positive Side Effects 0.21 0.15 0 1 Any Context-Specific Side Effect 0.86 0.35 0 1 % of Context-Specific Side Effects 0.24 0.19 0 1 Note: The different types of variables are grouped in Panels A through F, with the number of sessions and users for whom that information is available listed in parentheses. For each variable in each panel, we report the mean, standard deviation, minimum, and maximum values for that variable. Our variables are all dichotomous {0, 1} with the exception of THC, CBD, and the % of side effects variables, which are percentages range from 0 to 1, and the symptom relief measures, which are measured on a 0 to 10 scale with 0 being no discernable symptom level and 10 extreme symptom severity. Symptom change is measured as the ending symptom minus the starting symptom level for a range between −10 and 9. Starting symptom levels are restricted to range from 1 to 10, while ending symptom levels range from 0 to 10. Nineteen positive, thirteen negative, and ten context-specific side effects were available for selection in Panel F.

Table2 shows the results from regressing Symptom Relief on flower characteristics with each column representing a separate regression. When entered separately, the analyses show that only the product characteristic CBD percentage has a statistically significant effect on symptom relief; each additional percentage point of CBD contents is associated with a decrease of −0.04 (p < 0.01) in symptom severity levels. Starting symptom levels are important determinants of patient symptom relief with coefficients implying that once we restrict the sample to only those patients who reported CBD and THC, patients with low starting symptom levels experience worsening insomnia symptoms with consumption of Cannabis unless they are consuming sufficiently high percentages of CBD. When all product characteristics are included jointly, we find that smoking from a pipe or using a vaporizer is associated with greater symptom relief than smoking joints, and suggestive evidence that higher CBD levels are associated with greater symptom relief even after controlling for other characteristics of the flower consumed. The coefficient on C. sativa versus hybrid strains is large in magnitude at 2.481, but is statistically significant at only the 0.1 level. Figure1 explores these relationships in greater detail in the raw data, depicting the mean THC and CBD percentages by level of symptom relief. The left hand (low symptom relief) side of the figure is likely fairly noisy due to small session counts, but the figure does suggest that non-linearities may Medicines 2018, 5, 75 5 of 10 exist in the general improvement in symptom relief with higher CBD levels and lower THC levels. Figure1 also shows that THC potencies tend to be much higher than CBD potencies across all levels of symptom relief. This may indicate an interaction effect or that the optimal ranges may differ for the two cannabinoids.

Table 2. Effects of Product Characteristics on Symptom Relief—Regression Results.

Variable (1) (2) (3) (4) Panel A: Subtypes, omitted category = hybrid C. indica −0.227 0.176 (0.214) (0.220) C. sativa −0.214 2.481 * (0.492) (1.445) Panel B: Combustion Method, omitted category = joint Pipe −0.715 −1.686 ** (0.563) (0.758) Vape −0.823 −1.560 ** (0.583) (0.782) Panel C: THC and CBD THC (%) −4.759 −4.280 (2.978) (3.761) CBD (%) −3.828 *** −5.232 * (1.121) (2.841) Starting Symptom Level −0.763 *** −0.781 *** −0.951 *** −0.873 *** (0.058) (0.056) (0.095) (0.092) Constant 0.603 * 1.318 ** 2.599 *** 3.100 *** (0.362) (0.617) (0.920) (1.115) Observations 983 996 205 195 R-squared 0.341 0.337 0.562 0.613 Number of users 378 385 90 83 Notes: Each column represents a separate regression. Regressions control for individual user fixed effects. C. indica and C. sativa are relative to Hybrid, and Pipe and Vape are relative to Joint. Standard errors are clustered at the user level. ***Medicinesp < 0.01, 2018, ** 5, px FOR< 0.05, PEER * p REVIEW< 0.1. 6 of 10

30

25

20

15

% THC (%) 10 CBD (%)

5

0

-5 Symptom Relief

Figure 1. FigureTHC 1. and THC CBD and LevelsCBD Levels by by Symptom Symptom ReliefRelief Expe Experienced.rienced. Symptom Symptom relief is reliefmeasured ismeasured as as ending symptomending symptom level minus level minus starting starting symptom symptom levellevel with with −10− the10 maximum the maximum level of levelsymptom of symptomrelief. relief. Error barsError show bars the show range the of range values of values within within one standardone standard deviation deviation of of the the mean. mean. NoNo users who who reported reported THC and CBD percentages reported symptom levels worsening by more than one unit. The THC and CBD percentages reported symptom levels worsening by more than one unit. The number number of sessions for each level of symptom relief is reported in parentheses with the sample of sessionsrestricted for each to those level users of symptom reporting both relief CBD is an reportedd THC potency in parentheses levels, which withrange thebetween sample 0% and restricted to those users30% reporting and 2% and both 35%, CBD respectively. and THC potency levels, which range between 0% and 30% and 2% and 35%, respectively. In our regressions of side effects on product characteristics (Table 3), we find that product characteristics matter. In particular, C. sativa strains appear to be associated with more reporting of negative side effects, while vaping is associated with reduced reporting of negative side effects relative to consumption using joints and pipes. Individuals smoking joints appear to be the least likely to report positive or context-specific side effects. THC and CBD, however, do not appear to be associated with much variation in side effect reporting, except for a potential reduction in the extensive margin for context-specific side effects of 1.1 percentage points (p < 0.05). Higher starting symptoms are associated with increased negative side effects on both the extensive (any reported) and intensive (percent reported) margins as well as the reporting of a higher percent of positive side effects. Based on the R-squared, the product characteristics, individual user fixed effects and starting symptoms do a better job of explaining variation in positive and context-specific side effects than negative side effects.

Table 3. Effects of Product Characteristics on Side Effects—Regression Results.

(1) (2) (3) (4) (5) (6) Variable % of % of Context- % of Context- Negative Positive Negative Positive Specific Specific C. indica −0.034 0.001 0.037 −0.035 ** −0.004 −0.013 (0.026) (0.011) (0.036) (0.015) (0.039) (0.022) C. sativa 0.478 *** 0.105 *** −0.349 −0.002 −0.236 −0.018 (0.152) (0.028) (0.277) (0.088) (0.309) (0.131) Pipe −0.042 −0.044 0.025 0.116 ** 0.917 *** 0.252 *** (0.122) (0.038) (0.022) (0.050) (0.075) (0.050) Vape −0.484 *** −0.067 * 0.034 0.135 *** 0.993 *** 0.248 ***

Medicines 2018, 5, 75 6 of 10

In our regressions of side effects on product characteristics (Table3), we find that product characteristics matter. In particular, C. sativa strains appear to be associated with more reporting of negative side effects, while vaping is associated with reduced reporting of negative side effects relative to consumption using joints and pipes. Individuals smoking joints appear to be the least likely to report positive or context-specific side effects. THC and CBD, however, do not appear to be associated with much variation in side effect reporting, except for a potential reduction in the extensive margin for context-specific side effects of 1.1 percentage points (p < 0.05). Higher starting symptoms are associated with increased negative side effects on both the extensive (any reported) and intensive (percent reported) margins as well as the reporting of a higher percent of positive side effects. Based on the R-squared, the product characteristics, individual user fixed effects and starting symptoms do a better job of explaining variation in positive and context-specific side effects than negative side effects.

Table 3. Effects of Product Characteristics on Side Effects—Regression Results.

(1) (2) (3) (4) (5) (6) Variable % of % of % of Negative Positive Context-Specific Negative Positive Context-Specific C. indica −0.034 0.001 0.037 −0.035 ** −0.004 −0.013 (0.026) (0.011) (0.036) (0.015) (0.039) (0.022) C. sativa 0.478 *** 0.105 *** −0.349 −0.002 −0.236 −0.018 (0.152) (0.028) (0.277) (0.088) (0.309) (0.131) Pipe −0.042 −0.044 0.025 0.116 ** 0.917 *** 0.252 *** (0.122) (0.038) (0.022) (0.050) (0.075) (0.050) Vape −0.484 *** −0.067 * 0.034 0.135 *** 0.993 *** 0.248 *** (0.144) (0.037) (0.026) (0.049) (0.083) (0.063) THC (%) −0.136 −0.002 0.031 −0.200 0.808 0.506 (0.810) (0.093) (0.077) (0.304) (0.586) (0.455) CBD (%) 0.525 0.239 0.163 −0.008 −1.101 ** −0.006 (1.263) (0.159) (0.142) (0.154) (0.496) (0.241) Starting Symptom Level 0.074 *** 0.012 *** −0.002 0.012 *** 0.030 0.004 (0.025) (0.004) (0.002) (0.004) (0.021) (0.009) Constant 0.388 0.035 0.955 *** 0.084 −0.297 * −0.105 (0.329) (0.055) (0.035) (0.092) (0.155) (0.096) Observations 170 170 170 170 170 170 R-squared 0.128 0.165 0.382 0.123 0.355 0.101 N Users 70 70 70 70 70 70 Notes: Each column represents a separate regression. Regressions control for individual user fixed effects. C. indica and C. sativa are relative to Hybrid and Pipe and Vape are relative to Joint. Standard errors are clustered at the user level. *** p < 0.01, ** p < 0.05, * p < 0.1.

4. Discussion Given how important quality sleep is for optimizing mental and physical wellbeing, it is alarming how pervasive sleep disturbances are throughout society [2–4]. The limited effectiveness and risk of undesirable and potentially dangerous side effects of conventional pharmaceutical sleep aids [6,7] result in nearly 50% dissatisfaction rates [5]. Hence, it is not surprising why people with sleep disturbances commonly report regular experimentation with multiple types of sleep aids [26], including alcohol and Cannabis. Our results showed that on average, Releaf AppTM users experienced a statistically and clinically significant improvement (−4.5 points on a 0–10 point scale) in perceived insomnia levels. However, products made with C. sativa were associated with less symptom relief and more negative side effects than products made from C. indica or hybrid plant subtypes. Use of pipes and vaporizers was associated with greater symptom relief and more positive and context-specific side effects as compared to the use of joints, while vaporization was also associated with lower negative effects. CBD potency levels were associated with greater symptom relief than were THC levels, but the cannabinoid contents were generally not associated with differential reported side effects. The current results are consistent with survey-based studies showing increasing reported usage of cannabis for treating insomnia in healthy people and patients with other primary health Medicines 2018, 5, 75 7 of 10 conditions [13–15], and a patient preference for high CBD products [13,18,27]. In comparison to conventional prescription pharmaceutical sleep aids, CBD is generally believed to be much safer and often is described as non-psychoactive [16]. Prescription sleep aids in contrast, namely antidepressants (e.g., trazodone, amitriptyline, and doxepin), benzodiazepines (e.g., diazepam and lorazepam), gamma-aminobutyric acid (GABA) medications (zolpidem and eszopiclone), and anti-psychotics (aripiprazole, olanzapine, quetiapine and ) are associated with significant clinical drawbacks [6] and heightened risk of morbidity [28–31]. The phytocannabinoid family of CBDs are known to differ from other cannabinoids such as THC in several ways, including having little affinity to CB1 receptors, serving as an antagonist to the effects of THC, and functioning as anti-inflammatory and immuno-suppressant agents [32,33]. Orexin antagonists, such as suvorexant [34], as well as nemorexant and , which are currently in phase 3 clinical trials, are all dual antagonists of the orexin OX1 and OX2 receptors. They are being investigated for their potential use in treating sleep disorders. OX1 and OX2 receptors regulate several functions that overlap with cannabinoids, such as pain, wakefulness, and sleep. Both receptor types can form homo- and heterodimers with one-another and with CB1 receptors; however, orexin potentiation of CB1 signaling may result from orexin-promoted 2-AG (2-arachidonoyl glycerol, a native ligand of CB1 cannabinoid receptors) production and not necessarily from orexin-CB1 heterodimerization [23,35]. The activation of OX1 and OX2 receptors each modulate the effects induced by cannabinoids in different ways [36]. Whereas THC is a partial agonist of the CB1 receptor, CBD has a very low affinity for the CB1 receptor and instead acts as an indirect antagonist [37]. However, the fact that our results did not seem to show a clear relationship between THC or CBD and symptom relief suggests that other cannabinoid chemical(s) (e.g., cannabinols) and terpenes could contribute to changes in sleep experiences. Cannabinoid and terpene profiles vary across strains and we did find that the most frequently used cannabis strains for insomnia treatment were quite distinct in their chemotypic characteristics, highlighting the range of products and associated interactions among sub-compounds across products used by patients even just within flower. Therapeutically, cannabis consumption may also alleviate primary symptoms such as pain and anxiety, which are associated with sleep disturbances [16]. Unfortunately, due to cannabis’ continued Schedule I status and associated barriers to conducting medical cannabis research [24], no practical, naturalistic investigations have been completed on how patient-managed phytocannabinoid consumption affects discrete mechanisms (e.g., ventrolateral preoptic nucleus activation, memory consolidation) and other basic characteristics (e.g., sleep stages, circadian rhythm) involved in normal and aberrant sleep patterns. Despite the novelty and practical implications of our findings, the observational nature of the research design had unavoidable drawbacks, most notably the absence of a comparison group, which could have resulted in overestimation of the effectiveness of cannabis if unsatisfied users chose not to use the Releaf AppTM, or underestimation of cannabis’ effectiveness if users choose not to use the app as a result of accomplished satisfaction with product choices and their effects. It is also possible that the Releaf AppTM affected how users experience cannabis’ effects, and future research will benefit from examining both the effectiveness and influence of using electronic technology for patient medication management and monitoring. Small sample sizes could have also led to under-powered analyses, i.e., that other product characteristics matter but our sample is too small to pick them up at standard levels of statistical precision. Our study was also limited in the amount of information obtained by users and did not include detailed demographic characteristics, pre-app experience using cannabis, other types of sleep therapies, or type of sleep disorder. Finally, the study was limited to the accuracy of the product characteristics displayed on labels of the products consumed in the study, and there is a common problem of inaccurate (e.g., inflated) labeling practices in the medical cannabis industry in the U.S. [38]. Notwithstanding these limitations, this is the first study to measure how fundamental properties of self-directed Cannabis flower consumption affect immediate symptom relief from insomnia within users’ natural environments. Although no U.S. state has legalized medical cannabis for the Medicines 2018, 5, 75 8 of 10 treatment of sleep disorders, our results show that consumption of Cannabis flower is associated with significant improvements in perceived insomnia with differential effectiveness and side effect profiles. The widespread apparent use of cannabis as a sleep aid underscores the importance of further medical research regarding its risk-benefit profile and the effectiveness of cannabis as a substitute for other substances, including alcohol, over-the-counter and prescription sleep aids, and scheduled medications (e.g., and sedatives) [14,39–41], many of which are used in part as sleep aids.

Supplementary Materials: The following are available online at http://www.mdpi.com/2305-6320/5/3/75/s1, Table S1: Categories and Frequency of Side Effect Reporting, Figure S1: Releaf AppTM User Interface. Author Contributions: J.M.V. and S.S.S. conceived the study. F.B., K.K., B.H. independently designed and developed the Releaf AppTM and server infrastructure as part of their effort to help create an education tool for medical cannabis patients. S.S.S. conducted the analyses. J.M.V., S.S.S., and J.P.D. drafted the manuscript. All authors contributed substantially to its intellectual content and revision. Funding: This research was funded by the University of New Mexico Medical Cannabis Research Fund, mcrf.unm.edu. Acknowledgments: We thank all the donors to the University of New Mexico Medical Cannabis Research Fund for making this research possible. Conflicts of Interest: F.B., K.K., and B.H. are employed by Morebetter Ltd. The authors report no other conflicts of interests.

References

1. Abraham, O.; Pu, J.; Schleiden, L.J.; Albert, S.M. Factors contributing to poor satisfaction with sleep and healthcare seeking behavior in older adults. Sleep Health 2017, 3, 43–48. [CrossRef][PubMed] 2. Centers for Disease Control. Available online: https://www.cdc.gov/sleep/data_statistics.html (accessed on 15 March 2018). 3. Ellis, J.G.; Perlis, M.L.; Neale, L.F.; Espie, C.A.; Bastien, C.H. The natural history of insomnia: Focus on prevalence and incidence of acute insomnia. J. Psychiatr. Res. 2012, 46, 1278–1285. [CrossRef][PubMed] 4. Ohayon, M.M. Epidemiology of insomnia: What we know and what we still need to learn. Sleep Med. Rev. 2002, 6, 97–111. [CrossRef][PubMed] 5. Chung, S.; Youn, S.; Yi, K.; Park, B.; Lee, S. Sleeping pill administration time and patient subjective satisfaction. J. Clin. Sleep Med. 2016, 12, 57–62. [CrossRef][PubMed] 6. Morin, C.M.; Benca, R. Chronic insomnia. Lancet 2012, 379, 1129–1141. [CrossRef] 7. Fitzgerald, T.; Vietri, J. Residual effects of sleep medications are commonly reported and associated with impaired patient-reported outcomes among insomnia patients in the United States. Sleep Disord. 2015, 2015, 607148. [CrossRef][PubMed] 8. Auld, F.; Maschauer, E.L.; Morrison, I.; Skene, D.J.; Riha, R.L. Evidence for the efficacy of melatonin in the treatment of primary adult sleep disorders. Sleep Med. Rev. 2017, 34, 10–22. [CrossRef][PubMed] 9. Bent, S.; Padula, A.; Moore, D.; Patterson, M.; Mehling, W. Valerian for Sleep: A Systematic Review and Meta-Analysis. Am. J. Med. 2006, 119, 1005–1012. [CrossRef][PubMed] 10. Pragst, F.; Herre, S.; Bakdash, A. Poisonings with diphenhydramine—A survey of 69 clinical and 55 death cases. Forensic Sci. Int. 2006, 161, 189–197. [CrossRef][PubMed] 11. Radovanovic, D.; Meier, P.J.; Guirguis, M.; Lorent, J.P.; Kupferscchmidt, H. Dose-dependent toxicity of diphenhydramine overdose. Hum. Exp. Toxicol. 2000, 19, 489–495. [CrossRef][PubMed] 12. Gray, S.L.; Anderson, M.L.; Dublin, S.; Hanlon, J.T.; Hubbard, R.; Walker, R.; Yu, O.; Crane, P.K.; Larson, E.B. Cumulative Use of Strong Anticholinergics and Incident Dementia A Prospective Cohort Study. JAMA Intern. Med. 2015, 175, 401–407. [CrossRef][PubMed] 13. Bonn-Miller, M.O.; Babson, K.A.; Vandrey, R. Using cannabis to help you sleep: Heightened frequency of medical cannabis use among those with PTSD. Drug Alcohol Depend. 2014, 136, 162–165. [CrossRef][PubMed] 14. Bradford, A.C.; Bradford, W.D. Medical marijuana laws reduce prescription medication use in Medicare Part D. Health Aff. 2016, 35, 1230–1236. [CrossRef][PubMed] 15. Piper, B.J. Substitution of medical cannabis for pharmaceutical agents for pain, anxiety, and sleep. J. Psychopharmacol. 2017, 31, 569–575. [CrossRef][PubMed] Medicines 2018, 5, 75 9 of 10

16. National Academies of Sciences, Engineering, and Medicine; Health and Medicine Division; Board on Population Health and Public Health Practice; Committee on the Health Effects of Marijuana. An Evidence Review and Research Agenda. In The Health Effects of Cannabis and Cannabinoids: The Current State of Evidence and Recommendations for Research; National Academies Press: Washington, DC, USA, 2017. 17. Babson, K.A. Cannabis, Cannabinoids, and Sleep: A Review of the Literature. Curr. Psychiatry Rep. 2017, 19, 23. [CrossRef][PubMed] 18. Chagas, M.H.N.; Eckeli, A.L.; Zuardi, A.W.; Pena-Pereira, M.A.; Sobreira-Neto, M.A.; Sobreira, E.T.; Camilo, M.R.; Bergamaschi, M.M.; Schenck, C.H.; Hallak, J.E.C.; et al. Cannabidiol can improve complex sleep-related behaviours associated with rapid eye movement sleep behaviour disorder in Parkinson’s disease patients: A case series. J. Clin. Pharm. Ther. 2014, 39, 564–566. [CrossRef][PubMed] 19. Feinberg, I.; Jones, R.; Walker, J.M.; Cavness, C.; March, J. Effects of high dosage delta-9-tetrahydrocannabinol on sleep patterns in man. Clin. Pharmacol. Ther. 1975, 17, 458–466. [CrossRef][PubMed] 20. Gates, P.J.; Albertella, L.; Copeland, J. The effects of cannabinoid administration on sleep: A systematic review of human studies. Sleep Med. Rev. 2014, 18, 477–487. [CrossRef][PubMed] 21. Nicholson, A.N.; Turner, C.; Stone, B.M.; Robson, P.J. Effect of Delta-9-tetrahydrocannabinol and cannabidiol on nocturnal sleep and early-morning behavior in young adults. J. Clin. Psychopharmacol. 2004, 24, 305–313. [CrossRef][PubMed] 22. Whiting, P.F.; Wolff, R.F.; Deshpande, S.; Nisio, M.D.; Duffy, S.; Hernandez, A.V.; Keurentjes, J.C.; Lang, S.; Misso, K.; Ryder, S.; et al. Cannabinoids for Medical Use: A Systematic Review and Meta-analysis. JAMA 2015, 313, 2456–2473. [CrossRef][PubMed] 23. Thompson, M.D.; Takeshi, S.; Rainero, I.; Maj, M.C.; Kukkonen, J.P. multimerization versus functional interactions: Neuropharmacological implications for and cannabinoid signalling and pharmacogenetics. Pharmaceuticals 2017, 10, 79. [CrossRef][PubMed] 24. Stith, S.S.; Vigil, J.M.V. Federal barriers to Cannabis research. Science 2016, 352, 1182. [CrossRef][PubMed] 25. Releaf App. Available online: https://Releaf®appcom/ (accessed on 18 April 2018). 26. Mintel. OTC Sleep Aids US, MARCH 2017. 2017. Available online: http://academic.mintel.com.libproxy. unm.edu/sinatra/oxygen_academic/attachment/id=793087&seq=1 (accessed on 15 March 2018). 27. Belendiuk, K.A.; Babson, K.A.; Vandrey, R.; Bonn-Miller, M.O. Cannabis types and cannabinoid concentration preference among sleep-disturbed medicinal cannabis users. Addict. Behav. 2015, 50, 178–181. [CrossRef] [PubMed] 28. Fontanella, C.A.; Campo, J.V.; Phillips, G.S.; Hiance-Steelesmith, D.L.; Sweeney, H.A.; Tam, K.; Lehrer, D.; Klein, R.; Hurst, M. Benzodiazepine use and risk of mortality among patients with schizophrenia: A retrospective longitudinal study. J. Clin. Psychiatry 2016, 77, 661–667. [CrossRef][PubMed] 29. Nakafero, G.; Sanders, R.D.; Nguyen-Van-Tam, J.S.; Myles, P.R. The association between benzodiazepines and influenza-like illness-related pneumonia and mortality: A survival analysis using UK Primary Care data. Pharmacoepidemiol. Drug Saf. 2016, 25, 1263–1273. [CrossRef][PubMed] 30. Weich, S.; Pearce, H.L.; Croft, P.; Singh, S.; Crome, I.; Bashford, J.; Frisher, M. Effect of anxiolytic and hypnotic drug prescriptions on mortality hazards: Retrospective cohort study. Bmj 2014.[CrossRef][PubMed] 31. Sledzi´nski,P.;´ Zeyland, J.; Słomski, R.; Nowak, A. The current state and future perspectives of cannabinoids in cancer biology. Cancer Med. 2018, 7, 765–775. [CrossRef][PubMed] 32. Burstein, S. Cannabidiol (CBD) and its analogs: A review of their effects on inflammation. Bioorg. Med. Chem. 2015, 23, 1377–1385. [CrossRef][PubMed] 33. McPartland, J.M.; Duncan, M.; Di Marzo, V.; Pertwee, R.G. Are cannabidiol and ∆9-tetrahydrocannabivarin negative modulators of the endocannabinoid system? A systematic review. Br. J. Pharmacol. 2015, 172, 737–753. [CrossRef][PubMed] 34. Norman, J.; Anderson, S. Novel class of medications, orexin receptor antagonists, in the treatment of insomnia—Critical appraisal of suvorexant. Nat. Sci. Sleep 2016, 8, 239–247. [PubMed] 35. Scammell, T.; Winrow, C. Orexin receptors: Pharmacology and therapeutic opportunities. Annu. Rev. Pharmacol. Toxicol. 2011, 51, 243–266. [CrossRef][PubMed] 36. Flores, Á.; Julià-Hernández, M.; Maldonado, R.; Berrendero, F.; Flores, Á.; Julià-Hernández, M. Involvement of the orexin/hypocretin system in the pharmacological effects induced by ∆(9)-tetrahydrocannabinol. Br. J. Pharmacol. 2016, 173, 1381–1392. [CrossRef][PubMed] Medicines 2018, 5, 75 10 of 10

37. Pertwee, R.G. The diverse CB1 and CB2 receptor pharmacology of three plant cannabinoids: ∆9-tetrahydrocannabinol, cannabidiol and ∆9-tetrahydrocannabivarin. Br. J. Pharmacol. 2008, 153, 199–215. [CrossRef][PubMed] 38. Bonn-Miller, M.O.; Loflin, M.J.E.; Thomas, B.F.; Marcu, J.P.; Hyke, T.; Vandrey, R. Labeling accuracy of cannabidiol extracts sold online. JAMA 2017, 318, 1708–1709. [CrossRef][PubMed] 39. Haroutounian, S.; Ratz, Y.; Ginosar, Y.; Furmanov, K.; Saifi, F.; Meidan, R.; Davidson, E. The effect of medicinal cannabis on pain and quality-of-life outcomes in chronic pain: A prospective open-label study. Clin. J. Pain 2016, 32, 1036–1043. [CrossRef][PubMed] 40. Stith, S.S.; Vigil, J.M.; Adams, I.M.; Reeve, A.P. Effects of legal access to cannabis on Scheduled II-V drug prescriptions. J. Am. Med. Dir. Assoc. 2018, 19, 59–64. [CrossRef][PubMed] 41. Vigil, J.M.; Stith, S.S.; Adams, I.M.; Reeve, A.P. Associations between medical cannabis and prescription opioid use in chronic pain patients: A preliminary cohort study. PLoS ONE 2017, 12, e0187795. [CrossRef] [PubMed]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). fphar-09-00916 August 27, 2018 Time: 10:23 # 1

ORIGINAL RESEARCH published: 28 August 2018 doi: 10.3389/fphar.2018.00916

Patient-Reported Symptom Relief Following Medical Cannabis Consumption

Sarah S. Stith1, Jacob M. Vigil2*, Franco Brockelman3, Keenan Keeling3 and Branden Hall3

1 Department of Economics, The University of New Mexico, Albuquerque, NM, United States, 2 Department of Psychology, The University of New Mexico, Albuquerque, NM, United States, 3 The MoreBetter Ltd., Washington, DC, United States

Background: The Releaf AppTM mobile software application (app) data was used to measure self-reported effectiveness and side effects of medical cannabis used under naturalistic conditions. TM Edited by: Methods: Between 5/03/2016 and 12/16/2017, 2,830 Releaf App users completed Marco Leonti, 13,638 individual sessions self-administering medical cannabis and indicated their Università degli Studi di Cagliari, Italy primary health symptom severity rating on an 11-point (0–10) visual analog scale in Reviewed by: real-time prior to and following cannabis consumption, along with experienced side Simona Pichini, Istituto Superiore di Sanità, Italy effects. Francesco Paolo Busardò, TM Sapienza Università di Roma, Italy Results: Releaf App responders used cannabis to treat myriad health symptoms, the *Correspondence: most frequent relating to pain, anxiety, and depressive conditions. Significant symptom Jacob M. Vigil severity reductions were reported for all the symptom categories, with mean reductions [email protected] between 2.8 and 4.6 points (ds ranged from 1.29–2.39, ps < 0.001). On average, higher

Specialty section: pre-dosing symptom levels were associated with greater reported symptom relief, and This article was submitted to users treating anxiety or depression-related symptoms reported significantly more relief Ethnopharmacology, (ps < 0.001) than users with pain symptoms. Of the 42 possible side effects, users a section of the journal Frontiers in Pharmacology were more likely to indicate and showed a stronger correlation between symptom relief Received: 11 April 2018 and experiences of positive (94% of sessions) or a context-specific side effects (76%), Accepted: 26 July 2018 whereas negative side effects (60%) were associated with lessened, yet still significant Published: 28 August 2018 symptom relief and were more common among patients treating a depressive symptom Citation: Stith SS, Vigil JM, Brockelman F, relative to patients treating anxiety and pain-related conditions. Keeling K and Hall B (2018) Conclusion: Patient-managed cannabis use is associated with clinically significant Patient-Reported Symptom Relief Following Medical Cannabis improvements in self-reported symptom relief for treating a wide range of health Consumption. conditions, along with frequent positive and negative side effects. Front. Pharmacol. 9:916. doi: 10.3389/fphar.2018.00916 Keywords: pain, anxiety, depression, cannabis, marijuana, quality of life, symptom management, side effects

Frontiers in Pharmacology| www.frontiersin.org 1 August 2018| Volume 9| Article 916 fphar-09-00916 August 27, 2018 Time: 10:23 # 2

Stith et al. Medical Cannabis Effects

INTRODUCTION users and 23,373 user interactions, we included only cannabis consumption sessions with reported starting symptom levels Medicinal cannabis use is expanding rapidly in the United States, greater than 0 (on a 0–10, 11-point scale) and ending symptom with thousands of new users daily, particularly older patients levels reported within 90 min of the start of the session, resulting and people with significant health concerns, treating many in a final sample of 2,830 users and 13,638 individual sessions different symptoms (Centers for Disease Control and Prevention, for analysis. The Releaf AppTM measures 27 possible negative 2016; Han et al., 2016). Most patients have a wide variety symptom categories and 42 possible side effects. Symptoms were of medicinal cannabis products available to them, ranging ultimately derived from qualifying conditions across medical from traditional flower to edibles and tinctures. Naturalistic cannabis programs in the United States, along with a few observational studies are generally well-suited for capturing suggested by dispensaries and patients. The side effects (called how patients manage their treatment decisions in real-life, and “feelings” within the app) were crowd-sourced among Releaf how patient-managed cannabis therapies may contribute to AppTM developers, beta testers, dispensaries, and patients, and symptom relief and potential side effects from use. Observational included 19 positive, 12 negative, and 11 context-specific side research designs allow patients to use the myriad Cannabis strains effects available for selection. Supplementary Tables S1, S2 in and cannabis-derived formulations (e.g., concentrates, tinctures, the Supplemental Appendix provide descriptive statistics for all edibles, topicals, suppositories, toothpaste) made at home and/or symptoms and side effects. commercially available and widely used in society, and can User sessions consist of a series of electronic instructions incorporate the breadth of health conditions for which medical for recording characteristics of the cannabis medication (e.g., cannabis has been sanctioned for use at the state-level. Lastly, strain, potency, formulation), pre-dosing symptom severity observational studies also circumvent research barriers associated rating along an 11-point visual analog facial pain scale from 0 (no with cannabis’ Schedule I status under United States federal law, detectable symptom level) to10 (severe), the timing of cannabis which makes randomized controlled trials (RCTs) challenging to consumption, a post-dosing symptom severity rating, and the conduct (Stith and Vigil, 2016; National Academies of Sciences, option to indicate any of the 42 listed side effects at any time Engineering, and Medicine, 2017). during the session. Among our primary sample of users, 2,332 Since its release in 2016, the commercially developed Releaf users reported side effects during 10,535 sessions. AppTM application (app; Releaf App, 2018) has been the only publically available, incentive-free patient educational software Study Outcomes program designed for recording how individual cannabis usage Our goal was to calculate changes in patient-perceived symptom sessions may correspond to immediate changes in primary severity, the prevalence of positive and negative side effects symptom intensity levels and experienced side effects. This associated with cannabis consumption, and whether the electronic assessment tool enables patients to monitor and reported-effects differs depending on the symptom for which manage their cannabis consumption decisions under naturalistic users were seeking treatment. We measured changes in conditions while avoiding the limitations of retrospective survey symptom relief by subtracting the ending symptom level from collection methods (e.g., memory bias, social desirability effects). the beginning symptom (possible range from −10 to 10). We used the Releaf AppTM repository of over 2,830 patients (Supplementary Figure S1 in the Supplemental Appendix and 13,368 individual cannabis administration sessions to provides a frequency table for each level of symptom relief.) examine two research questions: How does cannabis used Side effects were recorded as {0,1} variables for whether the under naturalistic conditions affect user-experienced symptom user selected that side effect from the menu. We categorize the relief and side effects? Does the magnitude of experienced side effects as positive, negative, or context-specific and then symptom relief and the prevalence of side effects vary across convert these categories of side effects into {0,1} outcomes, count symptom categories? The results have clinical relevance for outcomes and outcomes measuring the portion of total available understanding how patient-managed medical cannabis therapies side effects in that category a user selected. may correspond to changes in symptom intensity and potential side effects among people using cannabis for treating distinct Statistical Analysis health conditions (Hill and Weiss, 2016; Rubin, 2017). We use means comparisons and least squares regression models to estimate the absolute and relative symptom changes and side effect profiles resulting from the cannabis user sessions. We also MATERIALS AND METHODS created an adjusted symptom relief profile score, the mean change in symptom levels plus the absolute number of listed negative Study Design side effects, to provide a relative metric of cost-benefit tradeoffs A naturalistic observational research design, approved by the associated with cannabis use. Due to the small user counts for Institutional Review Board at the University of New Mexico, some of the reported symptoms, the large number of possible was used to analyze the Releaf AppTM user-submitted data symptoms, and to facilitate interpretation in our regression recorded between 5/03/2016 and 12/16/2017. Releaf AppTM is analysis, we aggregate the most commonly reported symptoms a cross-platform (iOS and Android) mobile and tablet app across three broad symptom categories that included: Anxiety backed by a secure cloud programming interface for capturing, Symptoms (agitation/irritability, anxiety, insomnia, stress, and processing, and storing anonymized user data. Out of 4,369 total muscle spasms), Pain Symptoms (ten pain categories), and

Frontiers in Pharmacology| www.frontiersin.org 2 August 2018| Volume 9| Article 916 fphar-09-00916 August 27, 2018 Time: 10:23 # 3

Stith et al. Medical Cannabis Effects

Depression Symptoms (depression). The remaining types of the prevalence of positive, negative, and context-specific side symptoms are less frequently reported or not clearly categorized. effects by the aggregated symptom categories (anxiety, pain, We also report the full regression results for the three and depression symptoms). For completeness, we include a fifth categories of side effects (positive, negative, and context-specific) column including the remaining discrete symptom categories and the sign for regressions of symptom relief on the full which did not fall under the three aggregated symptom range of 42 side effects. Standard errors are clustered at categories. Little variation exists in starting and ending symptom the user level to control for heteroskedasticity and arbitrary levels and the symptom relief experienced, with the average user correlation. reporting a symptom decrease of 3.7. With regards to side effects, those with depression have a higher probability of reporting negative or context-specific side effects. The most common RESULTS positive side effects are “relaxed” (64%), “peaceful” (54%), and “comfy” (38%), the most common negative side effects are “dry Figure 1 shows the starting and ending symptom severity levels, mouth” (23%), “foggy” (22%), and “forgetful” (13%) and the most the change in levels, the Cohen’s d of the difference, and the common context-specific side effects are “high” (32%), “sleepy” adjusted symptom relief profile score for each of the 27 discrete (27%), and “thirsty” (27%). symptom categories. For all symptoms, the null hypothesis that Table 2 examines how symptom relief varies across the the starting symptom severity level is less than or equal to the broader symptom categories, with the constant representing ending symptom severity can be rejected at the p < 0.001 level. the mean adjusted symptom change for the omitted category, Using the adjusted symptom relief measure (symptom relief plus (patients with pain-related symptoms). The first two regressions negative side effects), all but users with convulsions, dizziness, shown in Table 2 indicate that people with anxiety and depression excessive appetite, or tremors experienced a net improvement report greater relief from using cannabis than people with chronic in their symptom severity levels. Even for these symptoms, the pain, and users with higher starting symptom levels report adjusted mean symptom relief score still indicates a net benefit greater symptom relief. (The effects of cannabis on anxiety and from use and the lack of a statistically significant change likely depression symptoms are not statistically different from each relates more to the small number of observations rather than the other, although they are both greater than the effect of cannabis lack of an effect, given that these symptoms together constituted on pain-related symptoms). Negative responses or increases in less than 3% of users and less than 1% of our sample. For all other symptom severity do occur, but the intercept in combination with symptoms, the null hypothesis of an increase or no change in the the starting symptom level predicts that increases in symptom adjusted symptom relief score can be rejected at the p < 0.001 severity levels predominantly occur among users with starting level. symptoms equal to one. The third column in Table 2 shows that Table 1 provides additional information on starting and cannabis is more effective for anxiety and depression symptoms ending symptom severity levels, mean symptom relief, and than for pain-related symptoms among patients reporting higher

FIGURE 1 | Patient-reported symptom relief following medical cannabis consumption. Values in parantheses are the symptom category sample size, Cohen’s d, and adjusted symptom relief score (symptom relief + number of negative side effects), respectively.

Frontiers in Pharmacology| www.frontiersin.org 3 August 2018| Volume 9| Article 916 fphar-09-00916 August 27, 2018 Time: 10:23 # 4

Stith et al. Medical Cannabis Effects

TABLE 1 | Descriptive statistics – symptom levels and experienced side effects.

Overall Anxiety symptoms Pain symptoms Depression symptoms Other

N Sessions 13638 5343 4267 1440 2588 N Users 2830 1679 1223 577 1026 Starting symptom level 6.2 ± 2.2 6.2 ± 2.3 6.3 ± 2.0 6.5 ± 2.2 5.8 ± 2.4 Ending symptom level 2.5 ± 2.2 2.2 ± 2.2 3.0 ± 2.1 2.5 ± 2.2 2.4 ± 2.3 Symptom relief −3.7 ± 2.6 −4.0 ± 2.8 −3.3 ± 2.3 −4.0 ± 2.7 −3.4 ± 2.8 Better 94.2% 94.8% 94.7% 95.4% 91.6% Same 2.7% 2.4% 2.8% 2.4% 3.2% Worse 3.1% 2.8% 2.5% 2.2% 5.2% Any positive side effect 94.4% 94.7% 94.5% 93.9% 94.2% Any negative side effect 60.0% 60.0% 58.9% 65.5% 58.8% Any context-specific side effect 76.2% 75.2% 75.9% 80.1% 76.6% # of positive side effects 4.6 ± 3.2 4.6 ± 3.2 4.4 ± 3.1 4.8 ± 3.4 4.8 ± 3.4 # of negative side effects 1.4 ± 1.7 1.4 ± 1.7 1.3 ± 1.6 1.6 ± 1.9 1.3 ± 1.7 # of context-specific side effects 2.0 ± 1.9 2.0 ± 1.9 1.9 ± 1.9 2.1 ± 1.9 2.0 ± 1.9 % of positive side effects 24% 24% 23% 26% 25% % of negative side effects 11% 11% 10% 13% 10% % of context-specific side effects 20% 20% 19% 21% 20%

Symptoms designated as treatable with benzodiazepines (Anxiety Symptoms) include agitation/irritability, anxiety, insomnia, muscle spasms, and stress. Symptoms associated with Opioid treatment (Pain Symptoms) include all ten pain conditions. Depression is the only symptom designated as treatable with antidepressants.

TABLE 2 | Reported symptom relief for users treating anxiety, pain, and depression.

Outcome = symptom relief

(1) (2) (3)

Constant (opioid mean) −3.309∗∗∗ 1.120∗∗∗ 0.355∗∗ (−3.459 to −3.160) (0.804 to 1.436) (0.034 to 0.675) Anxiety symptoms −0.704∗∗∗ −0.763∗∗∗ 0.365∗ (−0.944 to −0.465) (−0.953 to −0.574) (−0.062 to 0.792) Depression symptoms −0.723∗∗∗ −0.563∗∗∗ 0.643∗ (−1.060 to −0.385) (−0.817 to −0.310) (−0.021 to 1.308) Starting symptom level (1–10) −0.706∗∗∗ −0.582∗∗∗ (−0.757 to −0.656) (−0.639 to −0.525) Anxiety∗start −0.181∗∗∗ (−0.259 to −0.102) Depression∗start −0.189∗∗∗ (−0.305 to −0.074) Observations 11,050 11,050 11,050 R2 0.018 0.372 0.377

Each column represents a separate regression. The omitted category is symptoms treatable with an opioid medication. Robust standard errors are clustered at the user level. The coefficients are reported in line with the variable names with confidence intervals below. Coefficients are reported with 95% Confidence Intervals below. ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.10.

symptom severity levels (A graphical representation of this appetite, insomnia, loss of appetite, nausea, gastrointestinal pain, relationship is presented in Supplementary Figure S2 in the stress, and tremors than they do for treating back pain. Patients Supplemental Appendix). reported less symptom relief for treating impulsivity, headache, In order to take advantage of the full range of symptom and nerve pain as compared to relief for treating back pain. The categories available to Releaf AppTM users, we also ran symptom relief for the other discrete symptom categories was regressions including dummy variables for each of the symptoms, indistinguishable from the reported symptom relief associated using back pain as the omitted category. After controlling for with back pain. starting symptom level, clustering the standard errors at the user Table 3 explores whether patients using cannabis to treat pain, level, and using a statistical significance threshold of p < 0.05, anxiety, or depressive symptoms differ in their experiences of our results indicate that patients report greater symptom relief positive, negative, or context-specific side effects. Chows tests for treating agitation/irritability, anxiety, depression, excessive (Chow, 1960) showed that users with anxiety-related symptoms

Frontiers in Pharmacology| www.frontiersin.org 4 August 2018| Volume 9| Article 916 fphar-09-00916 August 27, 2018 Time: 10:23 # 5

Stith et al. Medical Cannabis Effects

TABLE 3 | Differences in side effect profiles across symptom categories.

Outcome = side effect type

Positive Negative Context-specific

Any

Constant (opioid mean) 0.966∗∗∗ 0.496∗∗∗ 0.695∗∗∗ (0.942 to 0.989) (0.428 to 0.565) (0.637 to 0.753) Anxiety symptoms 0.001 0.013 −0.006 (−0.012 to 0.015) (−0.033 to 0.059) (−0.049 to 0.037) Depression symptoms −0.006 0.066∗∗ 0.042∗ (−0.029 to 0.017) (0.002 to 0.131) (−0.005 to 0.090) Starting symptom level −0.003∗ 0.015∗∗∗ 0.010∗∗ (−0.007 to 0.000) (0.007 to 0.024) (0.002 to 0.019)

Number

Constant (opioid mean) 4.583∗∗∗ 1.081∗∗∗ 1.652∗∗∗ (4.013 to 5.154) (0.768 to 1.395) (1.356 to 1.947) Anxiety symptoms 0.182 0.077 0.077 (−0.100 to 0.465) (−0.104 to 0.257) (−0.113 to 0.268) Depression symptoms 0.476∗ 0.324∗∗ 0.134 (−0.010 to 0.962) (0.053 to 0.596) (−0.187 to 0.454) Starting symptom level −0.035 0.036∗∗ 0.044∗∗ (−0.142 to 0.072) (0.000 to 0.072) (0.003 to 0.085)

Percent of possible

Constant (opioid mean) 0.241∗∗∗ 0.083∗∗∗ 0.165∗∗∗ (0.211 to 0.271) (0.059 to 0.107) (0.136 to 0.195) Anxiety symptoms 0.01 0.006 0.008 (−0.005 to 0.024) (−0.008 to 0.020) (−0.011 to 0.027) Depression symptoms 0.025∗ 0.025∗∗ 0.013 (−0.001 to 0.051) (0.004 to 0.046) (−0.019 to 0.045) Starting symptom level −0.002 0.003∗∗ 0.004∗∗ (−0.007 to 0.004) (0.000 to 0.006) (0.000 to 0.009)

The first panel uses {0,1} outcomes for the presence of side effects in each category, the second uses the count of side effects reported by category, and the third uses the number of reported side effects for each category divided by the total number of possible side effects a user could select in that category. Robust standard errors are clustered at the user level. Coefficients are reported with 95% Confidence Intervals below. ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.10.

are no more or less likely than those with pain symptoms to relief, highlighting the importance of adjusting for starting report any of the three categories of side effects. Individuals symptom severity level and side effect profiles when evaluating with depression, however, are more likely to report negative the overall effectiveness of cannabis as a treatment modality. and context-specific side effects than positive side effects. Higher starting symptom levels are also associated with more negative or context-specific side effect reporting and this relationship persists DISCUSSION whether the side effect profile is defined as any of the side effects from that category of side effects, the number of side effects by This is the largest observational study to measure immediate category, or the percent of possible side effects in a category. changes in patient-reported symptom severity ratings and Table 4 tests whether different types of side effects are experienced side effects in real-time from using cannabis under associated with differences in symptom relief. The results are naturalistic conditions. Building on previous research showing robust across specifications; reporting positive or context-specific that cannabis may be an effective substitute for opioids (Hurd, side effects is associated with greater symptom relief, while 2016; Vigil et al., 2017) and other classes of prescription reporting negative side effects is associated with less symptom medications (e.g., sedatives; Piper et al., 2017; Stith et al., relief. For example, based on Column (4), a person with a starting 2017), we provide evidence that cannabis is used to treat symptom level of 5 who reports 100% of negative side effects many different types of symptoms for which conventional would experience a 0.5 point increase in symptom severity on pharmaceutical medications are typically prescribed, and that the a 1–10 scale, whereas a similar user who does not report any magnitude of reported symptom relief and side effect profiles negative side effects would experience 2.2 points of symptom from using cannabis varies for people with different symptoms.

Frontiers in Pharmacology| www.frontiersin.org 5 August 2018| Volume 9| Article 916 fphar-09-00916 August 27, 2018 Time: 10:23 # 6

Stith et al. Medical Cannabis Effects

TABLE 4 | Association of positive, negative, and context-specific side effects with symptom relief.

Outcome = symptom relief

(1) (2) (3) (4)

Any {0,1} Percent of possible in category

Positive −1.100∗∗∗ −1.344∗∗∗ −2.345∗∗∗ −2.899∗∗∗ (−1.360 to −0.841) (−1.578 to −1.111) (−3.046 to −1.643) (−3.653 to −2.145) Negative 0.174∗∗ 0.336∗∗∗ 2.311∗∗∗ 2.772∗∗∗ (0.015 to 0.334) (0.192 to 0.480) (1.461 to 3.161) (2.045 to 3.498) Context-specific −0.339∗∗∗ −0.239∗∗∗ −0.781∗∗ −0.417 (−0.540 to −0.138) (−0.413 to −0.065) (−1.495 to −0.068) (−0.931 to 0.096) Starting symptom level −0.660∗∗∗ −0.666∗∗∗ (−0.710 to −0.610) (−0.724 to −0.608) Constant −2.307∗∗∗ 1.894∗∗∗ −3.098∗∗∗ 1.100∗∗∗ (−2.625 to −1.989) (1.441 to 2.348) (−3.372 to −2.824) (0.818 to 1.382) Observations 10,535 10,535 10,535 10,535 R2 0.015 0.349 0.036 0.376

The first two columns measure use the existence of each category of side effect as independent variables, while the second two columns use the percent of possible in each category of side effects. The second and fourth columns include the starting symptom level. In all four regressions, the outcome is the change in symptom severity. Robust standard errors are clustered at the user level. Coefficients are reported with 95% Confidence Intervals below. ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.10.

The Releaf AppTM users consumed cannabis to treat a wide cannabis consumption. This pattern of responses could have been range of health symptoms, the most frequent relating to pain, a function of characteristics of the software user interface (e.g., anxiety, or depression. Clinically and statistically significant symptom intensity scale range), manner in which responders reductions in patient-reported symptom severity levels existed interacted with their mobile device (e.g., visual attention to in every single symptom category, suggesting that cannabis common symptom severity levels), or with the systemic nature may be an effective substitute for several classes of medications by which phytocannabinoids may affect the human mind and with potentially dangerous and uncomfortable side effects body. According to the endocannabinoid deficiency theory, and risky polypharmaceutical interactions, including opioids, many mental and physical health disturbances result from the benzodiazepines, and antidepressants (Weich et al., 2014; dysregulation of the body’s innate endocannabinoid system (ECS; Centers for Disease Control and Prevention, 2016; Fontanella Smith and Wagner, 2014; Di Marzo et al., 2015; Karhson et al., 2016; Rudd et al., 2016; Sharma et al., 2016). Higher et al., 2016; Russo, 2018), often described as a master network pre-dosing symptom levels were generally associated with greater of chemical signals that promote somatic and psychological post-dosing symptom relief and users treating an anxiety-related homeostasis, or psychobiological state-efficiency (Bermudez- symptom or depression showed stronger symptom relief than Silva et al., 2010; Silvestri and Di Marzo, 2013; Acharya et al., users treating a pain symptom, even though depression is not a 2017). The ECS consists of natural ligands (e.g., anandamide and condition approved for medical cannabis use in most states. 2-AG) and receptors (CB1 and CB2) that appear to play a major Similar to clinical reviews showing that cannabis is associated role in efficient regulation of a wide range of systems that include with numerous, yet generally non-serious side effects (Wang sleep, feeding (e.g., gut permeability and adipogenesis), libido and et al., 2008; Whiting et al., 2016), positive and context-specific fertility, pain perception, motivation, happiness, anxiety, learning side effects were more commonly reported than negative side and memory, social functioning, autoimmune responses, cellular effects by the Releaf AppTM users, with the most frequent redox, and cancer pathophysiology (Valvassori et al., 2009; reported side effects being positive (relaxed, peaceful, comfy) and Muccioli et al., 2010; Abdel-Salam et al., 2012; Cani, 2012; the least frequent side effects being negative (paranoid, confused, Burstein, 2015; Du Plessis et al., 2015; McPartland et al., 2015; headache). Positive side effect reporting was associated with the Karhson et al., 2016; Pava et al., 2016; Tegeder, 2016; Turcotte greatest reported symptom relief, followed by context-specific et al., 2016; Androvicova et al., 2017; Sierra et al., 2018). In other side effects, while negative side effects were associated with lower words, unlike conventional pharmaceutical approaches, which reported symptom relief. In general, patients treating depression largely target specific neurotransmitter sites (e.g., monoamine were more likely to indicate a negative side effect than patients neurotransmitter hypothesis; Delgado, 2000; Ng et al., 2015), treating anxiety- or pain-related symptoms, though even users cannabis may act to improve a broad spectrum of symptoms who reported only negative side effects reported significant by regulating homeostatic functioning, perhaps best described as decreases in moderate to severe symptom intensity levels after a system-modulating rather than symptom-modulating form of using cannabis. therapy. One of the most striking patterns in the current results was Notwithstanding the strengths of the naturalistic research the breadth of symptoms that appeared to improve following design and the potential implications of the study’s findings,

Frontiers in Pharmacology| www.frontiersin.org 6 August 2018| Volume 9| Article 916 fphar-09-00916 August 27, 2018 Time: 10:23 # 7

Stith et al. Medical Cannabis Effects

the study was limited primarily by the lack of a control flower vs. oils) can differ in their dose reliability, and researchers group, e.g., non-cannabis users with the same symptom using have offered guidelines for dosing titration and experimental a mobile device to indicate their immediate symptom intensity usage (Kahan et al., 2014; Pichini et al., 2018). However, until levels. There is also the potential confound of user-selection federal laws currently restricting pharmacodynamics research bias and exclusion of users that failed to complete sessions in the United States are reformed (Stith and Vigil, 2016) or even use the Releaf AppTM due to a lack of symptom investigators still have tremendous opportunities to develop and relief or negative side effects. (It is possible that selection bias incorporate innovative assessment tools, like the Releaf AppTM, could have worked in the opposite way, excluding patients into observational research designs for measuring how patients that are already satisfied with their cannabis choices and experience self-directed cannabis treatment in their normal therefore choose not to use the software app). This study everyday lives outside of clinical settings. chose to focus on the existence of symptom relief and side effects rather than offer clinical guidance as to which cannabis products offer preferential symptom relief and side effects AUTHOR CONTRIBUTIONS profiles. As such we did not include product characteristics, e.g., routes of administration, quantity and method of ingestion, JV and SS conceived the study. FB, KK, and BH independently and cannabinoid content, all of which are likely crucial for designed and developed the Releaf AppTM and server understanding how cannabis affects symptom relief and side infrastructure as part of their effort to help create an education effect manifestation. We only show that, on average, most tool for medical cannabis patients. SS conducted the analyses. cannabis users experience symptom relief. Future research JV and SS drafted the manuscript. All authors contributed will benefit by incorporating these contextual factors into substantially to its intellectual content and revision. measurements of patient decisions and by dissecting how fundamental characteristics of the cannabis products themselves affect immediate and longer term changes in symptom relief and FUNDING potential adverse consequences. Patients with certain health conditions such as neurological This research was supported in part by the University of New disorders (e.g., multiple sclerosis, seizures, epilepsy, headache) Mexico Medical Cannabis Research Fund (mcrf.unm.edu). may face differential risks for experiencing adverse effects or exacerbating their symptoms, for instance, depending on the amount of delta-9-tetrahydrocannabinol they consume, and ACKNOWLEDGMENTS caution should be used for patients considering using highly potent cannabis products (Solimini et al., 2017). Complicating All authors had access to the data in the study and take matters are the allogamous (variable) and unstable nature of the responsibility for the integrity of the data and the accuracy of the Cannabis plant and the inherent inconsistencies in the chemical data analyses. contents across plant batches and derived formulations, which are affected by genetic characteristics, but also environmental, cultivation, and storage conditions (Thomas and Pollard, 2016; SUPPLEMENTARY MATERIAL Pacifici et al., 2017, 2018). These factors present challenges for both medical cannabis consumers and researchers as patients The Supplementary Material for this article can be found never have continuous access to cannabis products with precisely online at: https://www.frontiersin.org/articles/10.3389/fphar. consistent chemotypes. Cannabis-based products (e.g., dried 2018.00916/full#supplementary-material

REFERENCES Burstein, S. (2015). Cannabidiol (CBD) and its analogs: a review of their effects on inflammation. Bioorg. Med. Chem. 23, 1377–1385. doi: 10.1016/j.bmc.2015. Abdel-Salam, O. M. E., El-Sayed El-Shamarka, M., Salem, N. A., El- 01.059 Din, M., and Gaafar, A. (2012). Effects of Cannabis sativa extract on Cani, P. D. (2012). Crosstalk between the gut microbiota and the endocannabinoid haloperidol-induced catalepsy and oxidative stress in the mice. EXCLI J. 11, system: impact on the gut barrier function and the adipose tissue. Clin. 45–58. Microbiol. Infect. 18, 50–53. doi: 10.1111/j.1469-0691.2012.03866.x Acharya, N., Penukonda, S., Shcheglova, T., Hagymasi, A. T., Basu, S., and Centers for Disease Control and Prevention (2016). Wide-Ranging Online Data for Srivastava, P. K. (2017). Endocannabinoid system acts as a regulator of immune Epidemiologic Research (WONDER). Atlanta, GA: CDC. homeostasis in the gut. Proc. Natl. Acad. Ssci. U.S.A. 114, 5005–5010. doi: Chow, G. C. (1960). Tests of equality between sets of coefficients in two linear 10.1073/pnas.1612177114 regressions. Econometrica 28, 591–605. doi: 10.2307/1910133 Androvicova, R., Horace, J., Stark, T., Drago, F., and Micale, V. (2017). Delgado, P. L. (2000). Depression: the case for a monoamine deficiency. J. Clin. Endocannabinoid system in sexual motivational processes: is it a novel Psychiatry 61(Suppl. 6), 7–11. therapeutic horizon? Pharmacol. Res. 115, 200–208. doi: 10.1016/j.phrs.2016. Di Marzo, V., Stella, N., and Zimmer, A. (2015). Endocannabinoid signaling 11.021 and the deteriorating brain. Nat. Rev. Neurosci. 16, 30–42. doi: 10.1038/nrn Bermudez-Silva, F. J., Viveros, M. P., McPartland, J. M., and Rodriguez de 3876 Fonseca, F. (2010). The endocannabinoid system, eating behavior and energy Du Plessis, S. S., Agarwal, A., and Syriac, A. (2015). Marijuana, phytocannabinoids, homeostasis: the end or a new beginning? Pharmacol. Biochemistry Behav. 95, the endocannabinoid system, and male fertility. J. Assist. Reprod. Genet. 32, 375–382. doi: 10.1016/j.pbb.2010.03.012 1575–1588. doi: 10.1007/s10815-015-0553-8

Frontiers in Pharmacology| www.frontiersin.org 7 August 2018| Volume 9| Article 916 fphar-09-00916 August 27, 2018 Time: 10:23 # 8

Stith et al. Medical Cannabis Effects

Fontanella, C. A., Campo, J. V., Phillips, G. S., Hiance-Steelesmith, bowel syndrome and other treatment-resistant conditions? Neuro Endocrinol. D. L., Sweeney, H. A., Tam, K., et al. (2016). Benzodiazepine use and Lett. 29, 192–200. risk of mortality among patients with schizophrenia: a retrospective Sharma, T., Guski, L. S., Freund, N., and Gøtzsche, P. C. (2016). Suicidality longitudinal study. J. Clin. Psychiatry 77, 661–667. doi: 10.4088/JCP.15m and aggression during antidepressant treatment: systematic review and meta- 10271 analyses based on clinical study reports. BMJ 352:i65. doi: 10.1136/bmj.i65 Han, B. H., Sherman, S., Mauro, P. M., Martins, S. S., Rotenberg, J., and Palamar, J. J. Sierra, S., Luquin, N., and Navarro-Otano, J. (2018). The endocannabinoid system (2016). Demographic trends among older cannabis users in the United States, in cardiovascular function: novel insights and clinical implications. Clin. Auton. 2006–13. Addiction 112, 516–525. doi: 10.1111/add.13670 Res. 1, 35–52. doi: 10.1007/s10286-017-0488-5 Hill, K. P., and Weiss, R. D. (2016). Minimal physical health risk associated Silvestri, C., and Di Marzo, V. (2013). The endocannabinoid system in energy with long-term cannabis use—but buyer beware. JAMA 315, 2338–2339. homeostasis and the etiopathology of metabolic disorders. Cell Metab. 17, doi: 10.1001/jama.2016.5181 475–490. doi: 10.1016/j.cmet.2013.03.001 Hurd, Y. L. (2016). Cannabidiol: swinging the marijuana pendulum from ‘weed’ Smith, S. C., and Wagner, M. S. (2014). Clinical endocannabinoid deficiency to medication to treat the opioid epidemic. Trends Neurosci. 40, 124–127. (CECD) revisited: can this concept explain the therapeutic benefits of cannabis doi: 10.1016/j.tins.2016.12.006 in migraine, fibromyalgia, irritable bowel syndrome and other treatment- Kahan, M., Srivastava, A., Spithoff, S., and Bromley, L. (2014). Prescribing smoked resistant conditions? Neuro Endocrinol. Lett. 35, 198–201. cannabis for chronic noncancer pain: preliminary recommendations. Can. Fam. Solimini, R., Rotolo, M. C., Pichini, S., and Pacifici, R. (2017). Neurological Physician 60, 1083–1090. disorders in medical use of cannabis: an update. CNS Neurol. Disord. Drug Karhson, D. S., Hardan, A. Y., and Parker, K. J. (2016). Endocannabinoid signaling Targets 16, 527–533. doi: 10.2174/1871527316666170413105421 in social functioning: an RDoC perspective. Transl. Psychiatry 6:e905. doi: 10. Stith, S. S., Vigil, J. M., Adams, I. M., and Reeve, A. P. (2017). Effects of legal access 1038/tp.2016.169 to cannabis on Scheduled II-V Drug Prescriptions. J. Am. Med. Dir. Assoc. 19, McPartland, J. M., Duncan, M., Di Marzo, V., and Pertwee, R. G. (2015). 59–64. doi: 10.1016/j.jamda.2017.07.017 Are cannabidiol and 19-tetrahydrocannabivarin negative modulators of the Stith, S. S., and Vigil, J. M. V. (2016). Federal barriers to Cannabis research. Science endocannabinoid system? A systematic review. Br. J. Pharmacol. 172, 737–753. 352:1182. doi: 10.1126/science.aaf7450 doi: 10.1111/bph.12944 Tegeder, I. (2016). Endocannabinoids as guardians of metastasis. Int. J. Mol. Sci. Muccioli, G. G., Naslain, D., Bäckhed, F., Reigstad, C. S., Lambert, D. M., Delzenne, 17:230. doi: 10.3390/ijms17020230 N. M., et al. (2010). The endocannabinoid system links gut microbiota to Thomas, B. F., and Pollard, G. T. (2016). Preparation and distribution of cannabis adipogenesis. Mol. Syst. Biol. 6:392. doi: 10.1038/msb.2010.46 and cannabis-derived dosage formulations for investigational and therapeutic National Academies of Sciences, Engineering, and Medicine (2017). Health and use in the United States. Front. Pharmacol. 7:285. doi: 10.3389/fphar.2016.00285 Medicine Division; Board on Population Health and Public Health Practice; Turcotte, C., Blanchet, M., Laviolette, M., and Flamand, N. (2016). The CB2 Committee on the Health Effects of Marijuana: An Evidence Review and Research receptor and its role as a regulator of inflammation. Cell. Mol. Life Sci. 73, Agenda. Washington, DC: National Academies Press. 4449–4470. doi: 10.1007/s00018-016-2300-4 Ng, J., Papandreou, A., Heales, S., and Kurian, M. (2015). Monoamine Valvassori, S. S., Elias, G., de Souza, B., Petronilho, F., Dal-Pizzol, F., Kapczinski, F., neurotransmitter disorders - Clinical advances and future perspectives. Nat. et al. (2009). Effects of cannabidiol on amphetamine-induced oxidative stress Rev. Neurol. 11, 567–584. doi: 10.1038/nrneurol.2015.172 generation in an animal model of mania. J. Psychopharmacol. 25, 274–280. Pacifici, R., Marchei, E., Salvatore, F., Guandalini, L., Busardò, F. P., and Pichini, S. doi: 10.1177/0269881109106925 (2017). Evaluation of cannabinoids concentration and stability in standardized Vigil, J. M., Stith, S. S., Adams, I. M., and Reeve, A. P. (2017). Associations preparations of cannabis tea and cannabis oil by ultra-high performance between medical cannabis and prescription opioid use in chronic pain patients: liquid chromatography tandem mass spectrometry. Clin. Chem. Lab. Med. 55, a preliminary cohort study. PLoS One 12:e0187795. doi: 10.1371/journal.pone. 1555–1563. doi: 10.1515/cclm-2016-1060 0187795 Pacifici, R., Marchei, E., Salvatore, F., Guandalini, L., Busardò, F. P., and Pichini, S. Wang, T., Collet, J. P., Shapiro, S., and Ware, M. A. (2008). Adverse effects of (2018). Evaluation of long-term stability of cannabinoids in standardized medical cannabinoids: a systematic review. Can. Med. Assoc. J. 178, 1669–1678. preparations of cannabis flowering tops and cannabis oil by ultra-high- doi: 10.1503/cmaj.071178 performance liquid chromatography tandem mass spectrometry. Clin. Chem. Weich, S., Pearce, H. L., Croft, P., Singh, S., Crome, I., Bashford, J., et al. (2014). Lab. Med. 28, 94–96. doi: 10.1515/cclm-2017-0758 Effect of anxiolytic and hypnotic drug prescriptions on mortality hazards: Pava, M. J., Makriyannis, A., and Lovinger, D. M. (2016). Endocannabinoid retrospective cohort study. BMJ 348:g1996. doi: 10.1136/bmj.g1996 signaling regulates sleep stability. PLoS One 11:e0152473. doi: 10.1371/journal. Whiting, P. F., Wolff, R. F., Deshpande, S., Di Nisio, M., Duffy, S., Hernandez, pone.0152473 A. V., et al. (2016). Cannabinoids for medical use: a systematic review and Pichini, S., Pacifici, R., Busardò, F. P., Tagliabracci, A., and Giorgetti, R. (2018). meta-analysis. JAMA 3913, 2456–2473. The challenge of clinical application of FM2 cannabis oil produced in Italy for the treatment of neuropathic pain. Eur. Rev. Med. Pharmacol. Sci. 22, 863–865. Conflict of Interest Statement: The authors FB, KK, and BH were employed by doi: 10.26355/eurrev_201802_14363 company MoreBetter Ltd. Piper, B. J., DeKeuster, R. M., Beals, M. L., Cobb, C. M., Burchman, C. A., Perkinson, L., et al. (2017). Substitution of medical cannabis for pharmaceutical The remaining authors declare that the research was conducted in the absence of agents for pain, anxiety, and sleep. J. Psychopharmacol. 31, 569–575. doi: 10. any commercial or financial relationships that could be construed as a potential 1177/0269881117699616 conflict of interest. Releaf App (2018). Available at: https://releafapp.com/ [accessed April 18, 2018]. Rubin, R. (2017). Medical marijuana is legal in most states, but physicians have little Copyright © 2018 Stith, Vigil, Brockelman, Keeling and Hall. This is an open-access evidence to guide them. JAMA 317, 1611–1613. doi: 10.1001/jama.2017.0813 article distributed under the terms of the Creative Commons Attribution License Rudd, R. A., Seth, P., David, F., and Scholl, L. (2016). Increases in drug and opioid- (CC BY). The use, distribution or reproduction in other forums is permitted, provided involved overdose deaths—United States, 2010-2015. MMWR Morb. Mortal. the original author(s) and the copyright owner(s) are credited and that the original Wkly. Rep. 65, 1445–1452. doi: 10.15585/mmwr.mm655051e1 publication in this journal is cited, in accordance with accepted academic practice. Russo, E. (2018). Clinical Endocannabinoid Deficiency (CECD): can this concept No use, distribution or reproduction is permitted which does not comply with these explain therapeutic benefits of cannabis in migraine, fibromyalgia, irritable terms.

Frontiers in Pharmacology| www.frontiersin.org 8 August 2018| Volume 9| Article 916 CHEMISTRY & BIODIVERSITY – Vol. 4 (2007) 1729

REVIEW

Cannabis, Pain, and Sleep: Lessons from Therapeutic Clinical Trials of Sativex, a Cannabis-Based Medicine

by Ethan B. Russo*a)b), Geoffrey W. Guya), and Philip J. Robsona) a) GW Pharmaceuticals, Porton Down Science Park, Salisbury, Wiltshire SP4OJQ, U.K. b) GW Pharmaceuticals, 20402 81st Avenue SW, Vashon, WA 98070, USA (phone: þ1-206-408-7082; fax: þ1-866-234-7757; e-mail: [email protected])

Cannabis sativa L. has been utilized for treatment of pain and sleep disorders since ancient times. This review examines modern studies on effects of D9-tetrahydrocannabinol (THC) and cannabidiol (CBD) on sleep. It goes on to report new information on the effects on sleep in the context of medical treatment of neuropathic pain and symptoms of multiple sclerosis, employing standardized oromucosal cannabis-based medicines containing primarily THC, CBD, or a 1 :1 combination of the two (Sativex). Sleep-laboratory results indicate a mild activating effect of CBD, and slight residual sedation with THC- predominant extracts. Experience to date with Sativex in numerous Phase I–III studies in 2000 subjects with 1000 patient years of exposure demonstrate marked improvement in subjective sleep parameters in patients with a wide variety of pain conditions including multiple sclerosis, peripheral neuropathic pain, intractable cancer pain, and rheumatoid arthritis, with an acceptable adverse event profile. No tolerance to the benefit of Sativex on pain or sleep, nor need for dosage increases have been noted in safety extension studies of up to four years, wherein 40–50% of subjects attained good or very good sleep quality, a key source of disability in chronic pain syndromes that may contribute to patients quality of life.

Introduction. – Sleep disorders are important syndromes in modern medicine that include parasomnias, or sleep-disruptive events, as well common associated afflictions such as snoring and sleep apnea. The most common disorder is insomnia, or lack of sleep, said by the National Institute of Neurological Disorders and Stroke to afflict 60million Americans [1]. Insomnia is a major risk factor for associated morbidity even in the absence of illness, and is associated with markedly increased prevalence of depression, anxiety, absenteeism [2], accidents [3], and utilization of health care resources [4]. Sleep disruption itself, as in shift work in nurses, may contribute notably to anxiety and functional bowel disorders [5]. When such sleep disturbances occur secondary to pain, they can be termed symptomatic insomnia. Pain at night at least three times a week was identified as a significant factor in excessive daytime sleepiness in older adults [6]. When sleep disturbance accompanies chronic pain or disease, attendant treatment becomes increasingly problematic. Despite the prevalence and pervasiveness of symptomatic insomnia, very few studies have addressed it, particularly with respect to possible effects of analgesics on sleep. For example, use of non-steroidal anti-inflammatory drugs may be associated with gastroesophageal reflux [7] that itself aggravates insomnia [8].

 2007 Verlag Helvetica Chimica Acta AG, Zürich 1730 CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)

Fewer studies yet have employed modern methods of electroencephalography (EEG) or polysomnography to assess sleep disorders associated with chronic pain. Results of recent investigations are sobering, as formal sleep monitoring of patients with advanced cancers demonstrated that opioid treatment and pain disrupted nocturnal sleep, prolonged sleep latency, and limited attainment of sleep stages 3 and 4 as well as rapid eye movement sleep [9]. Further investigation indicated that such sleep disturbances were attributable to opioid treatment itself, which contributed to depression and even enhanced pain [10]. In light of such data, it is clear that new approaches to chronic pain and resultant sleep disorder are necessary. Cannabis sativa L. and its derivatives have been known since ancient times for their analgesic, soporific, and hypnotic effects. While mentioned frequently as beneficial to sleep in a variety of pathological conditions in 19th-century sources on Indian hemp [11], modern studies on cannabinoids and their therapeutic effects on sleep have received little attention in modern medical literature until the last few years. As will be noted, these are indications for which standardized cannabis-based medicine promises palliation and symptomatic relief that may contribute greatly to patients global impressions and subjective sense of relief of their condition. The primary psychoactive ingredient of cannabis is D9-tetrahydrocannabinol

(THC), many of whose actions are mediated via the CB1 G-protein coupled receptors that cluster in nociceptive areas of the brain [12], spinal cord [13][14], and peripheral nervous system [15] (see [16] for an excellent review). THC Activity mimics that of the natural endocannabinoids, anandamide (AEA, arachidonylethanolamide) and 2- arachidonylglycerol (2-AG), that are likewise partial agonists on the CB1 receptor, that modulate pain responses in integrative centers such as the periaqueductal grey matter [17] and pain in relation to stress [18]. Another important phytocannabinoid, the non-psychoactive cannabidiol (CBD), is not only an analgesic, anti-inflammatory, and antioxidant in its own right [19–21], but it is also reported to allay various THC adverse effects including sedation, tachycardia, and anxiety [22]. Recent work has demonstrated that CBD antagonizes tissue necrosis factor alpha (TNF-a) in a rodent model of rheumatoid arthritis [23], and enhances adenosine receptor A2A signaling via inhibition of an adenosine transporter [24], suggesting an important therapeutic role in various inflammatory and chronic pain states. Additional cannabis components including and also have analgesic properties that may be significant [25]. Historical and scientific aspects of cannabinoids and pain have been described for migraine [26], obstetrics and gynecology [27] , and gastroenterological conditions [28]. A clinical endocannabinoid deficiency has been hypothesized in relation to migraine, fibromyalgia, and idiopathic bowel syndrome [29]. In the current review, we will examine modern studies on effects of THC and CBD on sleep, and then report new information on the effects of cannabis-based medicines on sleep as a secondary outcome measure in the context of randomized clinical trials of medical treatment of chronic pain states, including neuropathic pain (NP), symptoms of multiple sclerosis (MS), and rheumatoid arthritis.

Clinical Studies of Cannabinoids and Sleep. – The soporific qualities of cannabis were noted in the ancient Indian ayurveda tradition [30]. Subsequently, the great CHEMISTRY & BIODIVERSITY – Vol. 4 (2007) 1731 taxonomist Linnaeus recognized cannabis as narcotica and anodyna in his Materia Medica in the 18th century [31] (p. 214). William B. OShaughnessy reintroduced cannabis to Western medicine from India in the 19th century [32], wherein it produced sleep and pain reduction for victims of rheumatism and many other conditions. Benefits on sleep were noted in various pain states [11] throughout the 19th and early 20th century, when cannabis medicines subsequently fell from common medical usage due to lack of standardization and daunting problems with dosing and quality control. Scientific study of cannabinoids entered the modern era in the early 1960s with the isolation of THC [33]. Early studies revealed that THC reduced sleep latency in normal and insomniac subjects, and caused some suppression of slow wave sleep (Stages 3 and 4) [34], often with a residual hangover effect the next day [35]. No formal studies of cannabinoids to date have included electroencephalography or polysomnography in symptomatic conditions or chronic pain states. In a recent case report [36], treatment with Marinol (dronabinol, synthetic THC) effectively reversed serious insomnia in three patients afflicted with intractable pruritus associated with cholestatic liver disease. Similarly, in a limited trial of Marinol, 2.5 mg at night in five dementia patients, a reduction was observed in nocturnal motor activity (p¼0.028) [37]. A series of experiments with cannabidiol performed in Brazil were summarized in 1981 [38], with observations based on subjective sleep assessments. Of two subjects taking CBD 300 mg twice a day (BID) for 2 d, one reported having slept more heavily, but no performance abnormalities were evident. Ten more subjects took 200 mg CBD vs. placebo on four separate occasions with no significant differences in subjective functioning, or level of alertness. Two of four subjects taking CBD, 10mg BID for 20d, complained of isolated episodes of daytime somnolence on rare occasions. Another experiment compared placebo to CBD, 3 mg/kg/d divided BID in eight subjects. One reported somnolence for a week, another for the entire 30d, and a third reported improvement in baseline insomnia. Subsequently, this group assessed 15 subjects with 40, 80, and 160 mg oral doses of CBD as a hypnotic vs. nitrazepam, 5 mg, and placebo in a double-blind randomized trial. This low dose of benzodiazepine and lower dose of CBD produced little effect on sleep. The highest CBD dose, however, seemed to extend sleep and reduce episodic wakening in 10/15 subjects subjectively, while also reducing dream recall. No hangover symptoms were noted.

Cannabinoid Effects on Brain Chemistry in Sleep. – The key role of the endogenous cannabinoid system in regulation of sleep–wake cycles was suggested by the finding that the CB1 antagonist/inverse agonist SR 141716A produces arousal in rats at the expense of slow-wave sleep [39]. This was further highlighted by the finding that the endocannabinoid anandamide (AEA) seems to mediate sleep induction and interacts with oleamide in this regard [40]. Subsequently, a Japanese group demonstrated the inhibition of serotonin and ketanserin (5-HT2A antagonist) binding to the 5-Treceptors by AEA [41]. A mild but similar response has recently been demonstrated for CBD

[42], and cannabis terpenoids [43], suggesting a possible synergy with the CB1 agonist, THC. Certain components of cannabis are sedating in their own right (reviewed in [44], particularly terpineol [45]). 1732 CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)

Recently, CBD was shown to inhibit uptake of AEA, and weakly inhibit its hydrolysis [46], making it, in effect, an inducer of AEA function, and suggesting a modulatory role for this agent in sleep. Additionally, a functional role for endocanna- binoids in regulation of respiratory stability in sleep to prevent sleep apnea has been suggested [47]. Finally, it has recently been demonstrated that CBD administered intracerebroventricularly in rats increased wakefulness in the lights-on period, and increased enhancement of c-FOS expression in hypothalamus and dorsal raphe nucleus [48], supporting a clinical alerting effect for this agent [22], as discussed below.

New Data on Sleep Modulation with Cannabis-Based Medicine Extracts (CBMs). – GW Pharmaceuticals received a license from the British Home Office in 1998 to cultivate cannabis and extract it as a standardized botanical drug substance for formulation into finished pharmaceutical products. Early indications have focused on multiple sclerosis (MS) and chronic pain, especially neuropathic, or associated with cancer and rheumatoid arthritis. Chemovars of cannabis were selected via Mendelian genetics to express one predominant phytocannabinoid [49][50]. Cloned plants undergo liquid CO2 extraction to produce botanical drug substances that contain predominantly THC (Tetranabinex), CBD (Nabidiolex), or a 1 :1 combination of the two (Sativex ; Fig. 1) [51][52]. Sativex is administered oromucosally via a pump-action spray with each 100-ml pump-action actuation providing 2.7 mg of THC, 2.5 mg of CBD plus other phytocannabinoids, terpenoids, and phytosterols [25], in a base of 50% EtOH and 50% propylene glycol with 0.05% peppermint flavoring. Pharmacokinetic data on this material is available from recent publications [53]. The preparation has onset of activity in 15–40min, which allows patients to titrate dosing requirements according to pain levels or other symptoms with an acceptable profile of adverse events.

Fig. 1. Sativex oromucosal cannabis based medicine (photo: Ethan Russo,2003) CHEMISTRY & BIODIVERSITY – Vol. 4 (2007) 1733

A total of 1000 patient years of Sativex exposure in over 2000 experimental subjects has been amassed in Phase-II and -III clinical trials. A slight majority of subjects had no previous recreational or medicinal cannabis exposure, but comparative efficacy results have been identical in cannabis-experienced and cannabis-naïve cohorts with no evidence of inadequacy of subject blinding [54][55]. Patients are generally able to find a stable dose at which they obtain therapeutic relief without unwanted psychoactive effects. All randomized controlled trials (RCTs) were performed with Sativex added as an adjunct to existing drug regimens in patients with intractable symptoms, i.e., patients considered treatment-resistant and remained on best available analgesic therapy and hypnotic medication, if prescribed. A concerted effort has been made in this review to include data from all available Sativex clinical trials; no negative data were excluded. Sativex was approved in June 2005 for marketing as a prescription medicine in Canada under a Notice of Compliance with Conditions (NOC/c) for central neuro- pathic pain in multiple sclerosis (MS). An Investigational New Drug (IND) application to study Sativex in intractable cancer pain patients in the USA was approved by the FDA in January 2006. Two independent reviews of Sativex have recently been published [56][57]. The effects of oromucosal high-THC extract (Tetranabinex), 15 mg, and THC- CBD extract doses of 5 and 15 mg of THC-equivalent were assessed by Nicholson et al. in eight subjects with respect to nocturnal sleep, early morning performance, memory, and residual sleepiness in a double-blind placebo-controlled four-way cross-over study with EEG monitoring [58]. While the THC extract, 15 mg, alone produced little effect on sleep architecture, sleep latency was reduced, memory was impaired, and residual sleepiness and mood changes were observed (p<0.05). Both dose levels of combined THC-CBD extract decreased Stage 3 sleep (p<0.05) over placebo, and the 15-mg doses increased wakefulness (p<0.05) compared to 5-mg doses. The 5-mg doses of THC-CBD extract actually produced faster reaction times on the digit recall test (p< 0.05) over placebo. The authors noted that whereas impaired memory was observed the next day when 15-mg THC extract was given alone overnight, there were no such effects when THC was concomitantly accompanied by 15 mg of CBD, as in Sativex. Conclusions were that THC was sedative, while, in contrast, the presence of CBD was alerting, tended to counteract THC adverse effects on cognition, and impaired wakefulness. In subsequent Phase-II and -III clinical trials, sleep quality was assessed with questionnaires completed by clinical trial subjects. Visual Analogue Scales (VASs) and Numerical Rating Scales (NRSs) are familiar instruments to many clinicians and have traditionally been used to quantify patient-rated subjective experiences. The two types of scale have similar sensitivity and reliability, but NRS is generally preferred by patients for ease of use. NRS and VAS are well-established and validated for the measurement of pain [59]. As is the case with pain, there is no objective gold standard by which to quantify the quality and quantity of sleep in patients participating in clinical trials. For this reason, most of the studies included in this review utilized NRSs or VASs to measure sleep and sleep disturbance. For example, Wade et al. [60] used VASs attached to the following questions: How was your quality of sleep last night?/How much sleep did you get last night?/How did you feel when you awoke this morning? The anchors at each extremity of the 10-cm line were best imaginable and worst 1734 CHEMISTRY & BIODIVERSITY – Vol. 4 (2007) imaginable for the first two questions, and totally refreshedortotally unrefreshedfor the third. As an example of NRS, Rog et al. [61] used an 11-box (0–10) scale attached to the following instruction: On a scale of 0–10 please indicate how your nerve pain disrupted your sleep last night. Please tick one box only. The anchors were did not disrupt sleep and completely disrupts (unable to sleep due to pain). Such measures appear to have good face validity. These and other studies of cannabis-based medicines on pain and sleep are summarized in the Table. In a Phase-II study in 24 patients with intractable neurogenic symptoms including MS and chronic pain, Tetranabinex, Nabidiolex, and Sativex were tested in a double- blind-N-of-1 RCT vs. placebo by Wade et al. [66]. Significant improvement was seen with both Tetranabinex and Sativex on pain (especially neuropathic) (p<0.05), but post-hoc analysis showed symptom control was best with Sativex (p<0.0001), with slightly less intoxication than with THC-predominant extract. Sativex significantly improved sleep quality (p¼0.041; Study GWN19902; Fig. 2) [66]. The authors noted that, compared to placebo, the CBD-predominant extract significantly improved pain, the THC-predominant extract yielded significant improvements in pain, muscle spasm, spasticity, and appetite, and combined THC:CBD extracts (Sativex) significantly improved muscle spasm and sleep. They also observed that the visual analogue scale for Sativex was significantly improved over baseline for 20subjects in the sleep category (p<0.05). Of particular note in this trial was the confirmation of the CBD component as alerting, while high THC extract (Tetranabinex) improved sleep parameters (although not statistically significantly over placebo in this trial), while the combination of the two (Sativex) improved sleep synergistically.

Fig. 2. Compendium of results of Sativex on sleep in earlier Phase II–III RCTs in multiple sclerosis (MS) and intractable chronic pain

In a Phase-II double-blind crossover N-of-1 study of intractable chronic pain in 34 subjects by Notcutt et al. [67], visual analogue scales for pain were significantly improved for Tetranabinex and Sativex extracts over placebo (p<0.001). Sativex CHEMISTRY & BIODIVERSITY – Vol. 4 (2007) 1735 produced best results for pain in MS subjects (p<0.0042). Marked improvement was observed on sleep duration (p¼0.0001) and quality (p¼0.0001; Study GWN19901A; Fig. 2). The authors commented that Sativex, while having little effect on the recorded sleep hours, rather produced marked changes in reported sleep quality from the poor or fair to good categories. The sleep quality measure represented a global assessment by the subject of sleep duration, depth, and relative degree of sleep disruption. Finally, they posited that improvement of sleep by the drug might prove to be one of its major benefits in chronic pain and MS. In a Phase-III randomized placebo-controlled clinical trial in central neuropathic pain due to MS over 5 weeks in 66 patients by Rog et al., subjects showed mean NRS analgesia favoring Sativex over placebo (p¼0.009), and significant benefit of Sativex over placebo was observed in sleep disturbance (p¼0.003) [61] (Study GWMS0107; Fig. 2). In another Phase-III RCT in intractable pain in 79 subjects with MS, diabetic neuropathy, or other conditions by Notcutt et al. [72], the Sativex cohort utilized escape analgesia a mean of 20.57% of days vs. 50.12% for placebo (p¼0.002). Sleep disturbance was also reduced by Sativex vs. placebo (Study GWPS0105; Fig. 2), with a treatment difference favoring the former (p¼0.045). In a Phase-III double-blind placebo-controlled trial of peripheral neuropathic pain with allodynia in 125 subjects by Nurmikko et al.[73],Sativex produced highly statistically significant improvements in pain levels (p¼0.004) and dynamic allodynia (p¼0.042). Marked reductions in sleep disturbance were observed (p¼0.001; Study GWNP0101; Fig. 2)[73]. In the largest clinical study of brachial plexus avulsion and central neuropathic pain to date by Berman et al. [68] in 48 subjects in a double-blind cross-over design assessing oromucosal Tetranabinex, Sativex, and placebo, comparable benefits were noted in Box Scale-11 pain scores with Tetranabinex (p¼0.002) and Sativex extracts (p¼0.005). Sleep disturbance scores favored Sativex over placebo (p¼0.017) [68] (Study GWBP0101; Fig. 2), with sleep quality scores also favoring Sativex (p¼0.019). In another Phase-III RCT focusing on mixed neurogenic symptoms in MS by Wade et al. [60], the greatest improvement following Sativex was noted in spasticity (p¼ 0.001). Subjects also demonstrated benefit on sleep disturbance (p¼0.047; Study GWMS0001; Fig. 2). From this cohort, 137 patients elected to continue on Sativex in safety-extension (SAFEX) studies [74]. Rapid reductions were noted in the first twelve weeks in pain VAS in 47 affected patients with sustained improvements for more than one year. During that time, there was no escalation of dose indicating an absence of tolerance to analgesic or other therapeutic benefits of the preparation. Similarly, no withdrawal syndrome (as defined by Budney et al. [75]) was noted in a subset of 25 patients who voluntarily stopped the medicine abruptly. Upon resumption, benefits resumed at the prior established dosages. Improvements in sleep were also maintained [74]. Additional data from patients with central and peripheral neuropathic pain who completed these RCTs have been collected in a second SAFEX study of some 507 subjects taking Sativex for at least one, and up to four years. These data confirm the continued efficacy of Sativex in maintaining improvements in subjective sleep parameters. As in the prior SAFEX in MS subjects with mixed symptoms [74], no 1736 CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)

Table. Clinical Studies of Cannabis Based Medicines on Pain and Sleep

Drug Clinical indication Subject Trial duration Results/reference number (N) Cannabis HIV neuropathy 505 days >30% pain reduction vs. placebo (smoked) (p¼0.04), sleep NA [62] Cannador Spasticity in MS 419 15 weeks Improvement over placebo in subjective pain associated with spasm (p¼0.003), sleep (p¼0.025) [63] Cannador Post-herpetic 65 4 weeks No benefit observed on pain, neuralgia sleep NA [64] Cannador Post-operative pain 30Single doses, 1 day Decreasing pain intensity with each increasing dosage (p¼0.01). Sleep NA formally. One complaint of sleep disturbance [65] Sativex Neurogenic pain 20Series of 2-week Improvement with Tetranabinex N-of-1 and Sativex on VAS pain vs. placebo crossover blocks (p<0.05), symptom control best with Sativex (p<0.0001). Sativex improved sleep quality (p¼0.041) [66] Sativex Chronic intractable pain 24 12 weeks, series VAS pain improved over placebo of N-of-1 (p<0.001) especially in MS crossover blocks (p<0.0042). Sleep duration and quality both improved (p¼0.0001) [67] Sativex Brachial plexus avulsion 48 6 weeks in 3 Benefits noted in Box Scale-11 two-week pain scores with Tetranabinex crossover blocks (p¼0.002) and Sativex (p¼0.005) over placebo. Sativex improved sleep disturbance (p¼0.017) and sleep quality scores (p¼0.019) [68] Sativex Central neuropathic 66 5 weeks Numerical Rating Scale (NRS) pain in MS analgesia improved (p¼0.009), sleep disturbance (p¼0.003) vs. placebo [61] Sativex Peripheral neuropathic 125 5 weeks Improvements in NRS pain levels pain (p¼0.004), dynamic allodynia (p¼0.042), sleep disturbance (p¼0.001) vs. placebo [69] Sativex Rheumatoid arthritis 56 5 week Improvements over placebo morning pain on movement (p¼0.044), morning pain at rest (p¼0.018), DAS-28 (p¼ 0.002), and SF-MPQ pain at present (p¼0.016), sleep quality (p¼0.027) [70] CHEMISTRY & BIODIVERSITY – Vol. 4 (2007) 1737

Table (cont.) Drug Clinical indication Subject Trial duration Results/reference number (N) Sativex Pain after spinal injury 117 10days NSD in sleep disturbance and NRS pain scores, but improved Brief Pain Inventory (p¼0.032) and Patients Global Impression of Change (p¼0.001, odds ratio 3.4). Sativex Intractable cancer pain 177 2 weeks Improvements in NRS analgesia vs. placebo (p¼0.0142), Tetranabinex NSD. Sleep quality NSD [71] Sativex Intractable lower urinary 135 8 weeks Improvement in bladder severity tract symptoms in MS symptoms (p¼0.001) and nocturia episodes (p¼0.01) over placebo. dose escalation over time was necessary to maintain efficacy, supporting a lack of tolerance to this clinical benefit. Specifically, in an initial combined cohort of 287 subjects with central or peripheral neuropathic pain (Fig. 3), ca. 40% of subjects attained good-to-very-good sleep quality with maintenance of up to two years. Fewer than 20% of subjects had less than satisfactory results in their assessments of sleep quality.

Fig. 3. Cumulative data on sleep disturbances (sleep quality scores) in a long-term safety-extension (SAFEX) study of central and peripheral neuropathic pain patients treated with Sativex (Study GWEXT0102) 1738 CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)

An examination of adverse event profiles from the two SAFEX studies (137 and 537 subjects, resp.) reveals that complaints attributed to poor sleep or residual fatigue are infrequent after regular use of Sativex (Fig. 4).

Fig. 4. Graph of fatigue and other adverse events attributable to sedation or sleep disturbance in safety- extension studies of mixed symptoms of multiple selerosis (MS) (SAFEX GWMS0001, N¼137), and peripheral and central neuropathic pain (SAFEX GWNP0102, N¼507) taking Sativex for greater than one and up to four years. Rates of associated complaints are all less than 10%.

In a Phase-II double-blind, randomized placebo-controlled five-week study of 56 rheumatoid arthritis patients with Sativex by Blake et al. [70], employing nocturnal treatment only, subjects received a maximum of 6 sprays each evening (16.2 mg THCþ 15 mg CBD). In the final treatment week, many study measures favored Sativex over placebo: morning pain on movement (p¼0.044), morning pain at rest (p¼0.018), 28- joint disease activity score (DAS-28; p¼0.002), and Short Form McGill Pain Questionnaire (SF-MPQ) pain at present (p¼0.016). Sleep quality favored Sativex over placebo (p¼0.027) (Fig. 5,a). Results of a Phase-III study (N¼177) comparing Sativex, Tetranabinex, and placebo in intractable pain due to cancer unresponsive to opiates by Johnson and Potts [71] demonstrated that Sativex produced highly statistically significant improvements in analgesia (p¼0.0142), while Tetranabinex was not significantly different from placebo, suggesting that the presence of CBD in the Sativex preparation contributed to pain control. Sleep quality in this study was not significantly improved over placebo, perhaps due to its short duration of only three weeks. Similarly, in a Phase-II study of neuropathic pain after spinal injury, whereas no significant difference was noted in the primary outcome measure of average daily pain due to a large placebo response, the Brief Pain Inventory (BPI) did improve (p¼ 0.032), as did the Patients Global Impression (PGI) of change (p¼0.001, odds ratio 3.4). No changes in sleep over placebo were noted in this brief ten-day trial (unpublished findings). In a Phase-III RCT of MS patients with intractable lower urinary tract symptoms and frequent accompanying pain in 135 subjects, Sativex produced a significant CHEMISTRY & BIODIVERSITY – Vol. 4 (2007) 1739

Fig. 5. a) Effect of Sativex vs. placebo on rheumatoid arthritis sleep quality GWCRI016 (N¼58; 0–10 NRS, 0¼very good and 10¼very bad). b) Effect of Sativex vs. placebo on nocturia in intractable lower urinary tract symptom patients with multiple sclerosis (N¼135). improvement over placebo in bladder symptom severity (p¼0.001) and in nocturia episodes (p¼0.010) affecting sleep (Fig. 5,b)(Fowler et al., GW Pharmaceuticals data on file, manuscript in preparation). Common Adverse Events (AE) of Sativex acutely in RCTs have included complaints of bad taste, oral stinging, dry mouth, dizziness, headache, nausea, or fatigue, but do not generally necessitate discontinuation, and proved less common over time. Cumulative subject withdrawals from the RCTs secondary to AEs attributable to Sativex have occurred in 10.7% of all subjects, and in 10.8% of MS subjects (data on file, GW Pharmaceuticals, May 24, 2006). Figures ranged from 12.5% in the first Phase-II trial [60], while 0% of Sativex subjects withdrew due to attributable AEs in studies of lower urinary tract symptoms in MS [76] and brachial plexus avulsion [68]. Placebo-controlled trials have also been conducted with an oral plant-derived cannabis-based medicine, Cannador, which contains variable THC:CBD ratios [22]. This was examined in a large trial alongside Marinol (synthetic THC) and placebo in MS patients (Table). In neither the acute trial (CAMS) reported by Zajicek et al. [63], nor its 12-month long-term follow-up [77], were significant improvements noted in sleep with Cannador or Marinol. These data would support the proposition that benefits of cannabis-based medicines on sleep in the context of symptomatic treatment may be specific to a preparations formulation and/or delivery system, and the improvement with one preparation cannot necessarily be extrapolated to another. Another study of Cannador in post-operative pain (Table) showed decreased pain with increasing dosage, but sleep was not assessed formally (NA) [65]. One subject noted sleep disturbance. Finally, Cannador was utilized in a four-week study of post-herpetic neuralgia (Table), but no benefit was observed on pain, and sleep was NA formally [64]. 1740 CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)

Results are recently available from the first RCT of smoked cannabis on pain, in sensory neuropathy due to HIV/AIDS or its treatment (Table) [62]. A greater than 30% reduction in pain vs. placebo was noted in this five-day trial, but sleep effects were not reported. An additional study is planned in California to assess effects of cannabis on sleep disturbance in similarly affected patients (http://www.cmcr.ucsd.edu/geninfo/ drummond_abs.htm). The FDA has recently published guidelines for botanical medicines that mandate parameters required for New Drug Approval [78]. The difficulties inherent in standardizing herbal cannabis, and pulmonary issues associated with its inhalation [79], make it unlikely that regulatory approval would be attainable in most nations of the world [54]. No head-to-head trials of Sativex vs. smoked cannabis have been performed, but a comparison of AE profiles from self-selected SAFEX study subjects on Sativex with those of smoked-cannabis patients utilizing standardized cannabis in government programs in Canada [80] and the Netherlands [81][82] supports the concept that Sativex was much better tolerated, especially with respect to mental status and cognitive issues [54].

Discussion. – Chronic pain, neurological illness, and sleep disorders are clearly co- morbid conditions. Upwards of 80% of MS patients suffer from debilitating fatigue symptoms and complain of significant sleep disturbance. Additionally, chronic pain accompanies MS in up to 60% in some surveys, with a citation of 48% in a recent study [83], further compromising the ability of patients to attain rest. Tachibana et al. [84] noted that such problems in MS arise from legion sources: pain, spasticity, muscle spasm, restless legs syndrome, myoclonus, and lower urinary tract symptoms, resulting in sleep disturbance in 80% of 28 subjects. It was felt by these authors that these problems were rarely addressed therapeutically. MS may also be associated with sleep apnea, a condition that has recently been demonstrated to respond favorably to treatment with THC in an animal model [47]. A recent study of sleep and fatigue in 60MS subjects is quite germane [85], with over half noting difficulty with sleep disturbance at least two nights per week. Fatigue and excessive daytime sleepiness affected 64 and 32% of subjects, respectively. Those problems correlated best to difficulties with middle-of-the-night insomnia that subjects attributed most often to pain/discomfort (21.7%) or nocturia symptoms (72.5%). These symptoms were improved by Sativex treatment in the above discussed RCTs. Comparison of rates of fatigue, lethargy, somnolence, and insomnia in Sativex SAFEX subjects (Fig. 4) supports very remarkable amelioration compared to MS patients in the Stanton study [85], many of whom were already taking pharmacotherapy for such symptoms. The authors of the latter study specifically recommended symptomatic treatment of pain and nocturia as strategies to minimize sleep disturbance and its diurnal sequelae. Similar sleep complaints affect patients with other etiologies of neuropathic pain. A current review article strongly suggested treatment of chronic pain with agents that concomitantly improve sleep [86]. A survey of 173 adults with neuropathic pain reported significantly higher rates of sleep disturbance and daytime somnolence vs. controls [87], with improvement after institution of specific treatment. Unfortunately, sleep disturbance continued in 43% of 140subjects suffering from diabetic neuropathy despite treatment [88]. CHEMISTRY & BIODIVERSITY – Vol. 4 (2007) 1741

In a recent review [89], the authors state, The alterations of THC on sleep EEG and its rebound effect, its side effects before sleep induction, and its residual effects after awakening have contraindicated its clinical use as a sedative hypnotic. Data from the clinical research on Sativex reviewed in this article are not consistent with this conclusion. Rather, the available evidence to date would suggest that Sativex improvement in subjective sleep parameters, and satisfaction in patients with MS and neuropathic pain, with symptomatic relief of pain, spasms, nocturia, and related complaints. From limited sleep-laboratory information, it seems unlikely that its use will result in significant change in sleep architecture. Sativex does not benefit all patients, but in those who do respond, the beneficial effects are maintained consistently over time without evidence of tolerance, and are not accompanied by unusual cognitive sequelae [54]. Of course, additional in-depth studies are needed to confirm these contentions and might include formal neuropsychological testing and polysomnog- raphy. Sativex patients and their caregivers have remarked to their physicians how the medicine had transformed their lives through its ability to allow them more restful sleep, increase their daytime level of function, and markedly improve their quality of life. Its addition to the pharmacopoeia may be welcomed by patients, families, and physicians.

REFERENCES [1] Brain Basics: Understanding Sleep, National Institute of Neurological Disorders and Stroke, Bethesda, 2006. [2] G. K. Zammit, J. Weiner, N. Damato, G. P. Sillup, C. A. McMillan, Sleep 1999, 22, S379. [3] M. B. Balter, E. H. Uhlenhuth, J. Clin. Psychiat. 1992, 53 Suppl., 34; discussion 40–42. [4] R. M. Benca, J. Clin. Psychiat. 2001, 62 Suppl. 10, 33. [5] W. Z. Lu, K. A. Gwee, K. Y. Ho, Eur. J. Gastroenterol. Hepatol. 2006, 18, 623. [6] A. I. Pack, D. F. Dinges, P. R. Gehrman, B. Staley, F. M. Pack, G. Maislin, Ann. Neurol. 2006, 59, 893. [7] P. O. Katz, J. M. Scheiman, A. N. Barkun, Aliment. Pharmacol. Ther. 2006, 23 Suppl. 2,9. [8] D. A. Johnson, Rev. Gastroenterol. Disord. 2005, 5 Suppl, 2, S3. [9] K. P. Parker, D. L. Bliwise, S. Jain, J. Dalton, C. Vena, J. Clin. Oncol., 2005 ASCO Annual Meeting Proceedings 2005, 23, 8020. [10] K. P. Parker, D. L. Bliwise, J. Dalton, W. Harris, S. Jain, M. Kohles-Baker, M. Ribeiro, C. Vena, B. Viswanathan, J. Clin. Oncol., 2005 ASCO Annual Meeting Proceedings 2006, 24, 8526. [11] R. Mechoulam, in The Pharmacohistory of Cannabis sativa, Ed. R. Mechoulam, CRC Press, Boca Raton, 1986, p. 1–19. [12] J. D. Richardson, L. Aanonsen, K. M. Hargreaves, Eur. J. Pharmacol. 1997, 319, R3. [13] J. D. Richardson, L. Aanonsen, K. M. Hargreaves, J. Neurosci. 1998, 18, 451. [14] J. D. Richardson, L. Aanonsen, K. M. Hargreaves, Eur. J. Pharmacol. 1998, 345, 145. [15] J. D. Richardson, S. Kilo, K. M. Hargreaves, Pain 1998, 75, 111. [16] R. G. Pertwee, Prog. Neurobiol. 2001, 63, 569. [17] J. M. Walker, S. M. Huang, N. M. Strangman, K. Tsou, M. C. Sanudo-Pena, Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 12198. [18] A. G. Hohmann, R. L. Suplita, N. M. Bolton, M. H. Neely, D. Fegley, R. Mangieri, J. F. Krey, J. M. Walker, P. V. Holmes, J. D. Crystal, A. Duranti, A. Tontini, M. Mor, G. Tarzia, D. Piomelli, Nature 2005, 435, 1108. [19] R. G. Pertwee, in The pharmacology and therapeutic potential of cannabidiol, Ed. V. DiMarzo, Kluwer Academic, Dordrecht, 2004, p. 32–83. 1742 CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)

[20] R. G. Pertwee, in Cannabidiol as a potential medicine, Ed. R. Mechoulam, Birkhäuser, Basel, 2005, p. 47–65. [21] A. J. Hampson, M. Grimaldi, J. Axelrod, D. Wink, Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 8268. [22] E. B. Russo, G. W. Guy, Med. Hypotheses 2006, 66, 234. [23] A. M. Malfait, R. Gallily, P. F. Sumariwalla, A. S. Malik, E. Andreakos, R. Mechoulam, M. Feldmann, Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 9561. [24] E. J. Carrier, J. A. Auchampach, C. J. Hillard, Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 7895. [25] J. M. McPartland, E. B. Russo, J. Cannabis Ther. 2001, 1, 103. [26] E. B. Russo, J. Cannabis Ther. 2001, 1,21. [27] E. Russo, J. Cannabis Ther. 2002, 2,5. [28] R. G. Pertwee, Gut 2001, 48, 859. [29] E. B. Russo, Neuroendocrinol. Lett. 2004, 25,31. [30] E. B. Russo, in Cannabis in India: Ancient lore and modern medicine, Ed. R. Mechoulam, Birkhäuser, Basel, 2005, p. 1–22. [31] C. A. Linne´, Materia medica per regna tria naturae, Wolfgang Waltherum, Lipsiae et Erlangae, 1772. [32] W. B. OShaughnessy, Trans. Med. Phys. Soc. Bengal 1838–1840, 71; W. B. OShaughnessy, Trans. Med. Phys. Soc. Bengal 1838–1840, 421. [33] Y. Gaoni, R. Mechoulam, J. Am. Chem. Soc. 1964, 86, 1646. [34] R. T. Pivik, V. Zarcone, W. C. Dement, L. E. Hollister, Clin. Pharmacol. Ther. 1972, 13, 426. [35] K. Cousens, A. DiMascio, Psychopharmacologia 1973, 33, 355. [36] G. W. Neff, C. B. OBrien, K. R. Reddy, N. V. Bergasa, A. Regev, E. Molina, R. Amaro, M. J. Rodriguez, V. Chase, L. Jeffers, E. Schiff, Am. J. Gastroenterol. 2002, 97, 2117. [37] S. Walther, R. Mahlberg, U. Eichmann, D. Kunz, Psychopharmacology (Berlin) 2006, 185, 524. [38] E. A. Carlini, J. M. Cunha, J. Clin. Pharmacol. 1981, 21, 417S. [39] V. Santucci, J. J. Storme, P. Soubrie, G. Le Fur, Life Sci. 1996, 58, PL103. [40] R. Mechoulam, E. Fride, L. HanuÐ, T. Sheskin, T. Bisogno, V. Di Marzo, M. Bayewitch, Z. Vogel, Nature 1997, 389, 25. [41] T. Kimura, T. Ohta, K. Watanabe, H. Yoshimura, I. Yamamoto, Biol. Pharm. Bull. 1998, 21, 224. [42] E. B. Russo, A. Burnett, B. Hall, K. K. Parker, Neurochem. Res. 2005, 30, 1037. [43] E. B. Russo, C. M. Macarah, C. L. Todd, R. Medora, K. Parker, 41st Annual Meeting of the American Society of Pharmacognosy, Seattle, WA, 2000. [44] E. B. Russo,  Handbook of psychotropic herbs: A scientific analysis of herbal remedies for psychiatric conditions, Haworth Press, Binghamton, 2001. [45] G. Buchbauer, L. Jirovetz, W. Jager, C. Plank, H. Dietrich, J. Pharm. Sci. 1993, 82, 660. [46] T. Bisogno, L. HanuÐ, L. De Petrocellis, S. Tchilibon, D. E. Ponde, I. Brandi, A. S. Moriello, J. B. Davis, R. Mechoulam, V. Di Marzo, Br. J. Pharmacol. 2001, 134, 845. [47] D. W. Carley, S. Paviovic, M. Janelidze, M. Radulovacki, Sleep 2002, 25, 391. [48] E. Murillo-Rodriguez, D. Millan-Aldaco, M. Palomero-Rivero, R. Mechoulam, R. Drucker-Colin, FEBS Lett. 2006, 580, 4337. [49] E. P. de Meijer, M. Bagatta, A. Carboni, P. Crucitti, V. M. Moliterni, P. Ranalli, G. Mandolino, Genetics 2003, 163, 335. [50] E. de Meijer, in The breeding of cannabis cultivars for pharmaceutical end uses, Eds. G. W. Guy, B. A. Whittle, P. Robson, Pharmaceutical Press, London, 2004, p. 55–70. [51] E. B. Russo, J. Cannabis Ther. 2003, 3,1. [52] B. A. Whittle, G. W. Guy, in Development of cannabis-based medicines; risk, benefit and serendipity, Eds. G. W. Guy, B. A. Whittle, P. Robson, Pharmaceutical Press, London, 2004, p. 427–466. [53] G. W. Guy, P. Robson, J. Cannabis Ther. 2003, 3, 121. [54] E. B. Russo, in The Solution to the Medicinal Cannabis Problem, Ed. M. E. Schatman, Taylor & Francis, Boca Raton, 2006, p. 165–194. [55] S. Wright, in GWMS001 and GWMS0106: maintenance of blinding, GW Pharmaceuticals, London, 2005, p. 8. CHEMISTRY & BIODIVERSITY – Vol. 4 (2007) 1743

[56] M. P. Barnes, Expert Opin. Pharmacother. 2006, 7,607. [57] J. Pe´rez, Drugs Today 2006, 42, 495. [58] A. N. Nicholson, C. Turner, B. M. Stone, P. J. Robson, J. Clin. Psychopharmacol. 2004, 24, 305. [59] J. T. Farrar, J. P. Young Jr., L. LaMoreaux, J. L. Werth, R. M. Poole, Pain 2001, 94, 149. [60] D. T. Wade, P. Makela, P. Robson, H. House, C. Bateman, Mult. Scler. 2004, 10, 434. [61] D. J. Rog, T. Nurmiko, T. Friede, C. Young, Neurology 2005, 65, 812. [62] D. I. Abrams, C. A. Jay, S. B. Shade, H. Vizoso, H. Reda, S. Press, M. E. Kelly, M. C. Rowbotham, K. L. Petersen, Neurology 2007, 68, 515. [63] J. Zajicek, P. Fox, H. Sanders, D. Wright, J. Vickery, A. Nunn, A. Thompson, Lancet 2003, 362, 1517. [64] G. Ernst, C. Denke, M. Reif, M. Schnelle, H. Hagmeister, International Association for Cannabis as Medicine, 3rd Conference on Cannabinoids in Medicine, Leiden, The Netherlands, September 9– 10, 2005. [65] A. Holdcroft, M. Maze, C. Dore, S. Tebbs, S. Thompson, Anesthesiology 2006, 104, 1040. [66] D. T. Wade, P. Robson, H. House, P. Makela, J. Aram, Clin. Rehabil. 2003, 17, 18. [67] W. Notcutt, M. Price, R. Miller, S. Newport, C. Phillips, S. Simmonds, C. Sansom, Anaesthesia 2004, 59, 440. [68] J. S. Berman, C. Symonds, R. Birch, Pain 2004, 112, 299. [69] T. J. Nurmikko, M. G. Serpell, B. Hoggart, P. J. Toomey, B. J. Morlion, 57th Annual Meeting of the American Academy of Neurology, Miami Beach, FL, April 9–16, 2005. [70] D. Blake, P. Robson, M. G. Ho, R. W. Jubb, C. McCabe, Rheumatology 2006, 45,50. [71] J. R. Johnson, R. Potts, The 38th British Pain Society Annual Scientific Meeting, Edinburgh, Scottland, 2005. [72] W. G. Notcutt, M. Sharief, I. Mutiboko, C. Hawkes, J. Bolt, N. Sarantis, Eur. J. Pain 2007, in press. [73] T. J. Nurmikko, M. G. Serpell, B. Hoggart, P. J. Toomey, B. J. Morlion, Neurology 2005, 64, A374. [74] D. T. Wade, P. M. Makela, H. House, C. Bateman, P. J. Robson, Mult. Scler. 2006, 12, 639. [75] A. J. Budney, J. R. Hughes, B. A. Moore, R. Vandrey, Am. J. Psychiat. 2004, 161, 1967. [76] C. M. Brady, R. DasGupta, C. Dalton, O. J. Wiseman, K. J. Berkley, C. J. Fowler, Mult. Scler. 2004, 10, 425. [77] J. P. Zajicek, H. P. Sanders, D. E. Wright, P. J. Vickery, W. M. Ingram, S. M. Reilly, A. J. Nunn, L. J. Teare, P. J. Fox, A. J. Thompson, J. Neurol. Neurosurg. Psychiat. 2005, 76, 1664. [78] Guidance for industry: Botanical drug products, U.S. Depatment of Health and Human Services, Food and Drug Administration, 2004, p. 48. [79] D. P. Tashkin, Monaldi Arch. Chest Dis. 2005, 63, 93. [80] M. E. Lynch, J. Young, Symposium on the Cannabinoids, Clearwater, FL, 2005, p. 42. [81] A. F. C. Janse, N. S. Breekveldt-Postma, J. A. Erkens, R. M. C. Herings, in Medicinal gebruik van cannabis, PHARMO Instituut (Institute for Drug Outcomes Research), 2004, p. 51. [82] R. W. Gorter, M. Butorac, E. P. Cobian, W. van der Sluis, Neurology 2005, 64, 917. [83] N. Figved, G. Klevan, K. M. Myhr, S. Glad, H. Nyland, J. P. Larsen, E. Harboe, R. Omdal, D. Aarsland, Acta Psychiat. Scand. 2005, 112, 463. [84] N. Tachibana, R. S. Howard, N. P. Hirsch, D. H. Miller, I. F. Moseley, D. Fish, Eur. Neurol. 1994, 34, 320. [85] B. R. Stanton, F. Barnes, E. Silber, Mult. Scler. 2006, 12, 481. [86] M. D. Sullivan, J. P. Robinson, Phys. Med. Rehabil. Clin. N. Am. 2006, 17, 381, vi-vii. [87] R. D. Hays, S. A. Martin, A. M. Sesti, K. L. Spritzer, Sleep Med. 2005, 6,41. [88] T. Tolle, X. Xu, A. B. Sadosky, J. Diabetes Complications 2006, 20, 26. [89] N. Pace, H. C. Frick, K. Sutin, W. Manger, G. Hyman, G. Nahas, in The medical use of marihuana and THC in perspective, Eds. G. G. Nahas, K. M. Sutin, D. J. Harvey, S. Agurell, Humana Press, Totowa, 1999, p. 767–780.

Received April 2, 2007 FEBS Letters 580 (2006) 4337–4345

Cannabidiol, a constituent of Cannabis sativa, modulates sleep in rats

Eric Murillo-Rodrı´gueza,*, Diana Milla´n-Aldacoa, Marcela Palomero-Riveroa, Raphael Mechoulamb, Rene´ Drucker-Colı´na a Depto de Neurociencias, Instituto de Fisiologı´a Celular, Ciudad Universitaria, Circuito Interior, Universidad Nacional Auto´noma de Me´xico, Me´xico DF, CP 04510, Mexico b Department of Medicinal Chemistry and Natural Products, School of Pharmacy, Hebrew University, Jerusalem, Israel

Received 2 March 2006; revised 10 April 2006; accepted 17 April 2006

Available online 10 July 2006

Edited by Jesus Avila

1. Introduction Abstract D9-tetrahydrocannabinol (D9-THC) and cannabidiol (CBD) are two major constituents of Cannabis sativa. D9-THC modulates sleep, but no clear evidence on the role of CBD is Cannabis sativa preparations (marijuana, hashish, bhang available. In order to determine the effects of CBD on sleep, it and others) are the most widely used illicit drugs in the world was administered intracerebroventricular (icv) in a dose of [1]. D9-tetrahydrocannabinol (D9-THC) and cannabidiol 10 lg/5 ll at the beginning of either the lights-on or the lights- (CBD) are two major constituents of marijuana [2,3]. D9- off period. We found that CBD administered during the lights- THC is a psychoactive compound and produces stereotypical on period increased wakefulness (W) and decreased rapid eye behaviours [4]. The cannabinoid receptor CB1 is thought to movement sleep (REMS). No changes on sleep were observed be responsible for the majority of the effects in the central during the dark phase. Icv injections of CBD (10 lg/5 ll) in- nervous system (CNS) elicited by D9-THC, since it binds and duced an enhancement of c-Fos expression in waking-related activates this receptor [5]. brain areas such as hypothalamus and dorsal raphe nucleus (DRD). Microdialysis in unanesthetized rats was carried out to CBD does not bind to CB1 and is not psychoactive. The characterize the effects of icv administration of CBD (10 lg/ chemistry of CBD has been well explored [6–8] and some of 5 ll) on extracellular levels of dopamine (DA) within the nucleus its CNS pharmacological properties have been examined [9], accumbens. CBD induced an increase in DA release. Finally, in including its anticonvulsant, sedative and anxiolytic effects order to test if the waking properties of CBD could be blocked by [10–12]. For instance, Guimaraes and co-workers found that the sleep-inducing endocannabinoid anandamide (ANA), animals CBD (2.5–10 mg/kg) induced an anxiolytic-like effect evalu- received ANA (10 lg/2.5 ll, icv) followed 15 min later by CBD ated by the elevated plus maze assay whereas Lastres-Becker (10 lg/2.5 ll). Results showed that the waking properties of et al. have reported that CBD displays neuroprotection against CBD were not blocked by ANA. In conclusion, we found that 6-hydroxydopamine toxicity [13,14]. CBD modulates waking via activation of neurons in the hypo- While it is well established that D9-THC increases sleep [15– thalamus and DRD. Both regions are apparently involved in the generation of alertness. Also, CBD increases DA levels as 17], contradictory results on the effects of CBD have been pub- measured by microdialysis and HPLC procedures. Since CBD lished. For example, Monti et al. reported a reduction of sleep induces alertness, it might be of therapeutic value in sleep disor- by systemic administration of CBD [18]. On the other hand, ders such as excessive somnolence. Carlini and Cunha showed that CBD improved sleep in insom- 2006 Federation of European Biochemical Societies. Published niacs [19]. Recently, Nicholson et al. found that 15 mg of CBD by Elsevier B.V. All rights reserved. administered to young adults increased wakefulness (W) dur- ing sleeping time [20]. Keywords: Anandamide; Cannabinoid receptor; Insomnia; The mechanism of sleep modulation by CBD remains un- Dopamine clear. Presumably, it could include changes in dopamine (DA) levels, as the nigrostriatal dopaminergic system has been pointed out as an important element in the manifestations of the cannabinoid-induced behavioural alterations. Indeed the extracellular levels of DA are enhanced following D9-THC administration [21–23]. The effects of CBD on catecholamine levels are not known. As indicated above although CBD does not bind to the CB1 *Corresponding author. Fax: +52 55 5622 5607. cannabinoid receptor [24,25], it elicits numerous CNS-associ- E-mail address: [email protected] (E. Murillo-Rodrı´guez). ated effects, including the sleep–wake cycle. In view of the divergent effects reported so far we decided to reinvestigate this Abbreviations: ANA, anandamide; ACSF, artificial cerebrospinal fluid; D9-THC, D9-tetrahydrocannabinol; CBD, cannabidiol; DA, dopamine; phenomenon using techniques different from those reported DMH, dorsomedial hypothalamic nucleus; DRD, dorsal raphe nucl- previously. First we determined the pharmacological effects eus; FAAH, fatty acid amide hydrolase; HVA, homovanillic acid; icv, of unilateral intracerebroventricular (icv) administration of intracerebroventricular; MPO, medial preoptic nucleus; NA, nor- CBD on sleep in rats. Then we analysed the effects of CBD adrenaline; AcbC, nucleus accumbens; OEA, oleoylethanolamide; REMS, rapid eye movement sleep; 5-HT, serotonin; SWS, slow wave on c-Fos immunoreactivity followed by measurement of DA sleep; W, wakefulness; 5-HIAA, 5-hydroxy-indoleacetic acid; L-DOPA, extracellular levels collected from nucleus accumbens (AcbC) 3,4-dihydroxy-L-phenylalanine using microdialysis and HPLC. Finally, we looked into the

0014-5793/$32.00 2006 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.febslet.2006.04.102 4338 E. Murillo-Rodrı´guez et al. / FEBS Letters 580 (2006) 4337–4345 possibility that anandamide (ANA) could block the alertness during the lights-on period, we administered the treatments at the induced by CBD as we have previously reported that the endo- beginning of the lights-on period and right after this, samples were col- cannabinoid ANA increases sleep [26–28]. lected during 4 h.

2.7. Analysis of DA Immediately after collection samples were injected into a HPLC 2. Materials and methods (BAS) for DA analysis. Briefly, the mobile phase consisted of mono- chloroacetic acid (0.1 M), sodium octylsulfate (223 mM) and disodium ethylenediaminetetraacetate dihydrate (0.5 mM) with a flow rate of 2.1. Animals 80 ll/min. Separation was achieved by a BAS microbore column (bio- Male wistar rats (n = 50; 250–300 g) were housed at constant tem- phase octyl, 5 lm, 250 · 4.6 mm; BAS). Electrochemical detection was perature (21 ± 1 C) and under a controlled light–dark cycle (lights- performed via BAS 4C detector. Chromatographic data was recorded on: 07:00–19:00 h). Food and water were provided ad libitum. on a PC computer and peak heights of DA in microdialysis samples were compared to standards. DA peaks were initially identified by run- 2.2. Surgery, EEG/EMG electrodes and icv cannulae ning samples (also 10 ll) that contained different concentrations of Under deep anesthesia (acepromazine [0.75 mg/kg], [2.5 mg/ DA. All details of this procedure were reported by our group previ- kg], and ketamine [22 mg/kg, ip]) all animals (n = 12) were implanted ously [31]. for sleep studies with EEG and EMG electrodes and a cannula (23 gauge) was placed into one lateral ventricle (A = 0.8; L = 1.6; 2.8. Immunohistochemical study H = 3.6; [29]). All electrodes and cannulae were placed and secured At the end of the experiments, animals received either vehicle or CBD onto the skull using dental cement. These procedures have been re- (10 lg/5 ll, icv). The treatments were administered at the beginning of ported previously by our group [26,27]. After surgeries, all animals the lights-on period and the animals were sacrificed 1-h post-injection were placed into the sleep-recording chambers for habituation. with a lethal dose of . They were perfused transcardially with 0.9% saline solution followed by paraformaldehyde and then fol- 2.3. Surgery, microdialysis guide-cannulae lowed by 20% sucrose–0.1 M PBS for 48 h. The brains were cut (frozen A different group of rats was used for microdialysis study (n = 14). A sections, 30 lm, coronal) and collected in 1:5 serial orders. Immunohis- guide-cannula (IC guide. BioAnalytical Systems [BAS], West Lafay- tochemistry for c-Fos was done as described in detail by our group [32]. ette, IN, USA) was placed stereotaxically into the Accumbens nucleus, In addition to the c-Fos study, sections were lightly counterstained with core (AcbC; target coordinates: A = +1.2; L = 2.0; H = 7.0; [29]). The Neutral Red to allow better visualization of the cannula track. All stud- guide-cannulae was then fixed onto the skull with a thin layer of dental ies were conducted in accordance with the principles and procedures cement. After surgery, each animal was placed into the Microdialysis described in the National Institutes of Health Guide for the Care and Bowl to habituate to the experimental conditions. All animals were Use of Laboratory Animals. allowed to recover for at least 7 days after all surgeries. 2.9. Statistical analysis 2.4. Administrations The data are presented as means and standard errors. Student t test CBD was prepared in our lab as previously reported [25] and ANA was used to compare control and CBD groups in the first two experi- was purchased from Sigma (USA). All compounds were dissolved in ments and a P-value < 0.05 was considered statistically significant. In vehicle (PEG/saline; 5:95 v/v). In order to test the effects of CBD on the following experiments, statistical analysis was carried out by one- the sleep during the lights-on or lights-off period, rats were injected way analysis of variance (ANOVA). When a significant P-value was icv either with vehicle or CBD (10 lg/5 ll) at 07:00 h or at 19:00 h. found among the groups, the post-hoc Sheffe´ test was used to asses dif- For the combination experiment in which we examined the effect of ferences in two group comparisons (STATVIEW). ANA on the effects produced by CBD, additional groups of rats were used. Control group (n = 6) received vehicle. The following groups re- ceived either ANA (10 lg/5 ll; n = 6) or CBD (10 lg/5 ll; n = 6). The last group (n = 6) received ANA (10 lg/2.5 ll) and 15 min later CBD 3. Results (10 lg/2.5 ll). Injections were done at 07:00 h. All icv administrations were done slowly over 1 ll/min with the 3.1. Effects of CBD in sleep injector left in the target for an additional 15 s to ensure extrusion from We injected CBD (n =6;10lg/5 ll) or vehicle (n = 6) during the tip and to minimize distribution of treatments upwards on the can- the normal sleeping period (07:00–19:00 h, lights-on) of rats. nulae. After all injections, the cannula was withdrawn and the stylet was replaced. Right after microinjections, animals were attached to CBD markedly increased the total time spent in W the sleep-recording system. (DF = 10; t = 2.267; P < 0.05; Fig. 1A). We also found that CBD induced a significant diminution of REM sleep 2.5. Analyses of sleep recordings (DF = 10; t = 3.756; P < 0.05), whereas no statistical changes The EEG/EMG data recordings were scored manually and epochs were observed in SWS. As shown in Fig. 1B, CBD enhanced for W, slow wave sleep (SWS) and rapid eye movement sleep (REMS) W about 1 h after drug administration (DF = 10; t = 2.507; were measured as described previously [26,27]. The analysis was re- stricted to 4 h after injections since pharmacokinetic studies show that P < 0.05) as well as during the last hour (DF = 10; CBD is rapidly absorbed. On iv administration, CBD has a terminal t = 3.808; P < 0.005). The effects of CBD on SWS hour by half life of 9 h [30]. hour are displayed in Fig. 1C. We found that CBD also mod- ified this sleep stage during the second hour (DF = 10; 2.6. Microdialysis sampling procedures t = 2.672; P < 0.05) and the fourth hour (DF = 10; One week after surgery and a day before the experiment, a microdi- t = 2.126; P < 0.05). Analysis of REMS hour by hour after alysis probe (1 mm of length; polyacrylonitrile, MWCO = 30000 Da; administration of CBD is shown in Fig. 1D. CBD inhibited 340 lm OD; BAS) was inserted through the guide cannula into the tar- get structure at 7:00 h and the tissue were allowed to stabilize for 24 h. the time spent in REMS across 4 h of sleep recordings; how- During this period artificial cerebrospinal fluid (ACSF, composition: ever, statistical difference was found only at the fourth hour NaCl (147 mM), KCl (3 mM), CaCl (1.2 mM), MgCl (1.0 mM), pH (DF = 10; t = 3.709; P < 0.005). 7.2) was perfused through a FEP Teflon Tubing (0.65 mm On the other hand, CBD injected (n = 6) at the beginning of OD · 0.12 mm ID) continuously using a 2.5 ml gastight syringe. All procedures have been reported previously [28]. A syringe pump the lights-off period did not modify the total time of W (CMA/100) controlled the speed of perfusion of the ACSF (flow rate: (DF = 10; t = 0.546; P > 0.59), SWS (DF = 10; t = 0.212; 1 ll/min). Since we had found changes in sleep after CBD injection P > 0.83) and REMS (DF = 10; t = 1.25; P > 0.23) compared E. Murillo-Rodrı´guez et al. / FEBS Letters 580 (2006) 4337–4345 4339

Fig. 1. The icv administrations of either vehicle or CBD (10 lg/5 ll) were done at the start of the lights-on period, and right after injections, Fig. 2. The icv injection of CBD (10 lg/5 ll) at the start of the lights- the sleep was recorded during 4 h. The effects of CBD on the total time off period did not modify the total time (4 h of sleep recordings) of W, of W, SWS and REMS are shown in panel A. Hour-by-hour group SWS and REMS as shown in panel A. The time course of W (B), SWS averages of W (B); SWS (C) and REMS (D) after icv injection of either (C) and REMS (D) show no significant statistical differences (means ± S.E.M. of total time of recording [%]; vs control, P < 0.05). vehicle or CBD (means ± S.E.M. of total time of recording (%); * vs * control, P < 0.05). brain regions, such as lateral hypothalamus and dorsal raphe to control group (n =6; Fig. 2A). Analysis hour by hour nucleus (DRD). Some of the most striking changes in c-Fos of the pharmacodynamic effects of CBD on sleep stages are expression were found within specific hypothalamic areas shown in Fig. 2B–D, respectively, showing no significant implicated in the alertness control (Fig. 3). Neurons of hypo- changes. thalamus are active during W [33], and because all animals were mainly awake, it was not surprising that c-Fos-immuno- 3.2. c-Fos expression after CBD injection reactivity hypothalamic neurons were activated in rats that re- Treatment with CBD (10 lg/5 ll) consistently increased the ceived CBD. Following unilateral icv injection of CBD (10 lg/ pattern of neuronal activation marked by Fos expression in 5 ll), labelled c-Fos cells were observed in hypothalamic areas, 4340 E. Murillo-Rodrı´guez et al. / FEBS Letters 580 (2006) 4337–4345

Fig. 4. Immunohistochemical staining for c-Fos in the DRD of the rat. Panel A shows the expression of c-Fos from a control rat (vehicle). Panel B shows the immunostaining obtained from a CBD-treated animal (10 lg/5 ll icv). The treatments were administered at the start of the lights-on period and the animals were sacrificed 1 h post- injection. Abbreviations: Aq, aqueduct (Sylvius); DRD, dorsal raphe nucleus, dorsal part. Scale bar: 100 lm. Fig. 3. Immunohistochemical staining for c-Fos in the hypothalamus of the rat. Panel A shows the expression of c-Fos from a control rat (vehicle). Panel B shows the c-Fos immunostaining obtained from a CBD-treated animal (10 lg/5 ll icv). The treatments were administered levels in control rats in the first hour (DF = 12; t = 12.006; at the beginning of the lights-on period and the animals were sacrificed P < 0.0001; Fig. 6B) as well as in the second hour (DF = 12; 1 h post-injection. Abbreviations: 3V, 3rd ventricle. Scale bar: 100 lm. t = 5.387; P < 0.0002; Fig. 6B). CBD also induced enhancement in DA levels one hour after injection (DF = 12; t = 3.290; P < 0.05; Fig. 6C) and this ef- mainly in the medial preoptic nucleus (MPO) and dorsomedial fect remained in the second hour (DF = 12; t = 2.149; hypothalamic nucleus (DMH). Therefore, we found that CBD P < 0.05; Fig. 6C) whereas 3,4-dihydroxy-L-phenylalanine (L- induced an increase in c-FOS expression in MPO compared DOPA) extracellular levels decreased one hour after injection with the control group (Fig. 3). of CBD (DF = 12; t = 2.799; P < 0.01; Fig. 6D) as well as in On the other hand, in the brainstem, only neurons in the the second hour (DF = 12; t = 2.281; P < 0.05; Fig. 6D). DRD showed significant labelling pattern of Fos-immunoreac- Lastly, serotonin (5-HT) was increased 1 h post-CBD tivity after treatment with CBD (10 lg/5 ll). Representative administration (DF = 12; t = 16.741; P < 0.0001; Fig. 6E) examples of c-Fos immunoreactivity are shown in Fig. 4. All as well as 2 h post-CBD injection (DF = 12; t = 19.058; icv injections were made unilaterally. No differences among P < 0.0001; Fig. 6E). Levels of 5-hydroxy-indoleacetic acid control and CBD group were evident in other brainstem nu- (5-HIAA) were decreased 1 h after administration of CBD clei, including the pedunculopontino, laterodorsal tegmental (DF = 12; t = 6.074; P < 0.0001; Fig. 6F). Fig. 6G shows that nuclei and locus coeruleus. no effect was found on extracellular levels of homovanillic acid (HVA) after CBD injection. 3.3. Effects of CBD on DA levels Fig. 5 identifies the location of the cannula in the AcbC nu- 3.4. Effects of the sleep-inducing endocannabinoid ANA on the cleus from where the DA was measured. The variation in waking-inducing properties of CBD location of the cannula across the rats was as follows: ante- As found in our previous experiments [26–28], ANA sub- rior–posterior was Bregma +1.70 mm to Bregma 0.70 mm; stantially decreased waking and increased total sleep time. lateral was 1.5–2.0 mm and dorsal–ventral was 7.5–7.9 mm. Administration of ANA 15 min before CBD did not block Results showed that levels of noradrenaline (NA) were sig- either the wake-inducing effect caused by CBD or the diminu- nificant increased compared with a control group (n =7)in tion in SWS (F = 22.472; DF = 20; P < 0.0001; Fig. 7). Despite the first hour (DF = 12; t = 2.844; P < 0.01; Fig. 6A) and the evident sleep-inducing effect caused by ANA, the enhance- in the second hour (DF = 12; t = 2.137; P < 0.005; Fig. 6A) ment in W as well as the diminution in total sleep time was still post-injection of CBD (10 lg/5 ll; n = 7). After injection of present with the administration of CBD after 4 h of sleep CBD, epinephrine levels were also enhanced compared with recordings. E. Murillo-Rodrı´guez et al. / FEBS Letters 580 (2006) 4337–4345 4341

DMH contains neurons that are specifically active during W [38]. As shown in Fig. 4, a population of neurons in the DRD exhibited enhanced c-Fos-immunoreactivity after CBD micro- injection compared to controls. The DRD firing activity is higher in the waking state and decreases during sleep, being virtually absent in REM sleep [39–41]. These findings empha- size the hypothesis that one of the neurochemical mechanisms underlying the vigilance-promoting action of CBD could be re- lated to its ability to enhance the serotoninergic transmission. It has been reported recently that CBD displays 5-HT agonists [42], suggesting with this a potential activation of 5-HT recep- tors via CBD. Catecholamine levels collected from AcbC were monitored during 4 h post-injection of CBD (10 lg/5 ll, icv). We found an increase in the levels of NA, epinephrine, DA and 5-HT (Fig. 6A–C and E, respectively). On the other hand, the extra- cellular levels of L-DOPA and 5-HIAA were decreased (Fig. 6D and F, respectively). The enhancement in DA levels in our study is supported by the findings of McPartland and Russo [43]. We believe that waking induced by CBD may be associated with the increase in the release of DA since it is known that lesions of DA cell groups induce a reduction in arousal in rats [44] as well as in Parkinson’s disease patients [45,46]. The regulation of the sleep–wake cycle in mammals’ includes neuromodulators such as DA; for instance, administration of DA receptor agonists [47] or DA uptake inhibitors [48] induces an increase in wak- ing. The role of DA on alertness has been also tested in invertebrates including Drosophila [49]. On the other hand, the relative changes in the concentrations of DA have been demonstrated by Lena et al. [50]. Using the microdialysis Fig. 5. Histological verification of the microdialysis probe position. Panel A shows the schematic representation of the localization of the technique, the authors reported that extracellular levels of microdialysis probe in the AcbC represented by the black bar. DA were elevated during waking [50]. Finally, the electro- Stereotaxic coordinates, drawings and abbreviations were taken from physiological activity of DA neurons is higher during alert- the Paxinos and Watson [29] atlas. Microphotography of the track of ness as suggested by Lu et al. [51]. The evidence mentioned the microdialysis probe in the AcbC is shown in Panel B. Abbrevi- ations: AcbC: nucleus accumbens, core; CPu, caudate putamen. Scale above indicates that indeed DA plays an active role in the bar: 100 lm. promotion of waking [52]. In our study, we found that CBD increased W and enhanced the extracellular levels of DA collected from AcbC, finding which is supported by 4. Discussion McPartland and Russo [43]. A possible mechanism of this ef- fect remains to be elicited. However, we cannot rule out some The experiments described in this report were designed to hypothesis, including that CBD might be acting at different throw light on the pharmacological effects of CBD on sleep receptors. For instance, it has been suggested that CBD patterns of rats. Several physiological parameters were consid- may bind to the vanilloid receptor [24] or by modulating ered, such as the effects of CBD on the sleep–wake cycle, DA Ca2+ influx. Recently, Drysdale et al. reported that CBD ele- formation, c-Fos expression and interaction with ANA on vated intracellular Ca2+ concentrations in hippocampal cul- sleep. tures [53]. Thus, the present results provide scope for Under our conditions, icv administrations of CBD (10 lg/ speculation on the role of CBD on the mechanisms of release 5 ll) during the lights-on period increased waking, but de- of DA. creased REM sleep in rats (Fig. 1A). The alertness was ob- While AcbC is a good target for DA sampling it may not be served already after the first hour post-injection (Fig. 1B). a good option for 5-HT sampling. However data obtained Injection of CBD during the lights-off period did not modify from microdialysis study suggests that the increase in NA, epi- the sleep wake-cycle (Fig. 2A). nephrine and 5-HT levels could be an additional distant target We also found that CBD increased c-Fos expression as re- for CBD effects. ported by others [34]. The induction of Fos protein encoded While D9-THC and CBD cause some similar actions (albeit by the immediate early gene c-fos, is often used as a marker possibly based on different mechanisms) such as anti-inflam- of neural activation. The present results demonstrate that matory [54] and antiemetic effects [55], here we show a CBD induced changes in the Fos-immunoreactive cells in dichotomy in their effects on sleep patterns. D9-THC in- some waking-related brain areas, including hypothalamic nu- creases sleep [15–17], whereas CBD induces an opposite ef- clei (Fig. 3) as well as DRD (Fig. 4). The hypothalamic nuclei fect. Nicholson et al. have reported that CBD increases have been associated with waking [35–37]. For example, the waking in humans [20]. 4342 E. Murillo-Rodrı´guez et al. / FEBS Letters 580 (2006) 4337–4345

Fig. 6. Extracellular levels of catecholamines measured in the AcbC during a 4-h period after the administration of either vehicle or cannabidiol (CBD, 10 lg/5 ll icv). Noradrenaline (A), epinephrine (B), dopamine (C), L-DOPA (D), 5-HT (E), 5-HIAA (F) and HVA (G). The treatments were administered at the beginning of the lights-on period and right after this, samples were collected. Each point represents the means ± S.E.M. of pM (* vs control, P < 0.05).

Little is known about the molecular mechanism of CBD ac- our study we hypothesized an enhanced effect of ANA on sleep tions. Contrary to D9-THC, it has very little affinity for the following CBD injection. Surprisingly, CBD increased waking known cannabinoid receptors, CB1 and CB2 [24]. We assume even in presence of ANA. We do believe that CBD might be that the effects of CBD found in our study might be through inhibiting FAAH; this could lead an increase in endogenous action on a receptor not yet described. An alternative explana- ANA levels in the brain. However, the use of the URB597 (a tion involves the fatty acid amide hydrolase (FAAH) (the FAAH inhibitor) enhances OEA levels in the brain in higher intracellular enzyme that catalyzes the hydrolysis of endoge- rates than ANA as reported by Fegley et al. [58]. In our lab, nous cannabinoids ligand, ANA [56] and the anorexic lipid we have tested the physiological properties of URB597 as well oleoylethanolamide (OEA) [57]). In the last experiment of as OEA on sleep–wake cycle and have found a significant in- E. Murillo-Rodrı´guez et al. / FEBS Letters 580 (2006) 4337–4345 4343

cellular levels of NA, epinephrine, DA, and 5-HT were en- hanced after CBD injections. The wake-inducing properties of CBD were not blocked by the sleep-inducing endocannabi- noid ANA. Activation of specific arousal regions may underlie the W produced by CBD in people with somnolence. Future studies will be needed to address the question of whether vanil- loid receptors, Ca2+ influx or the interactions between CBD and CB1-signalling pathways contribute to the alertness and DA release actions of CBD.

Acknowledgements: This work was supported by FIDEICOMISO UNAM given to R.D-C. and UNAM/DGAPA/PAPIIT (IN208206- 2). The authors thank David Koblos for English style and composition corrections.

Fig. 7. Effects of icv administrations of either vehicle, cannabidiol References (CBD, 10 lg/5 ll), anandamide (ANA; 10 lg/5 ll), ANA (10 lg/2.5 ll) and 15 min later CBD (10 lg/2.5 ll; ANA + CBD) on the total time of [1] Grinspoon, L., Bakalar, J.B., Zimmer, L. and Morgan, J.P. (1997) W and total sleep time (TST). The treatments were administered at the Marijuana addiction. Science 277 (5327), 750–752. beginning of the lights-on period and right after this, sleep was [2] Mechoulam, R., Shani, A., Edery, H. and Grunfeld, Y. (1970) recorded during 4 h. Each point represents the means ± S.E.M. of total Chemical basis of hashish activity. Science 169 (945), 611–612. time of recording (%; * vs control, P < 0.05). [3] Mechoulam, R. (1996) Marihuana chemistry. Science 168 (936), 1159–1166. [4] Adams, I.B. and Martin, B.R. (1996) Cannabis: pharmacology crease in waking. The mechanism of this observation is like- and toxicology in animals and humans. Addiction 91 (11), 1585– 1614. wise unknown. It is highly possible that CBD inhibits FAAH [5] Ameri, A. (1999) The effects of cannabinoids on the brain. Prog. activity and this leads to an increase in the levels of OEA, Neurobiol. 58 (4), 315–348. which facilitates waking. [6] Mechoulam, R. and Hanus, L. (2002) Cannabidiol: an overview Our study suggests that CBD induces alertness via activation of some chemical and pharmacological aspects. Part I: chemical aspects. Chem. Phys. Lipids 121 (1–2), 35–43. of neurons in the hypothalamus and DRD. Additionally, levels [7] Grotenhermen, F. (2004) Pharmacology of cannabinoids. Neuro. of DA, NA, epinephrine as well as 5-HT were enhanced after Endocrinol. Lett. 25 (1–2), 14–23. CBD injections. These effects can be induced by a novel cellu- [8] McGilveray, I.J. (2005) Pharmacokinetics of cannabinoids. Pain lar pathway involved in the alertness caused by CBD. It seems Res. Manag. 10 (A), 15A–22A. that the enzymatic process involved in the formation of cate- [9] Fadda, P., Robinson, L., Fratta, W., Pertwee, R.G. and Riedel, G. (2004) Differential effects of THC- or CBD-rich cannabis cholamines might be under the influence of CBD. Likely the extracts on working memory in rats. Neuropharmacology 47 (8), component enzymes in the pathway could be inhibited or stim- 1170–1179. ulated by CBD. For example, we found that CBD increased [10] Chesher, G.B., Jackson, D.M. and Malor, R.M. (1975) Interac- tion of D9-tetrahydrocannabinol and cannabidiol with phenobar- DA but decreased levels of L-DOPA, meaning that the activity bitone in protecting mice from electrically induced convulsions. J. of tyrosine hydroxylase could be under inhibition whereas Pharm. Pharmacol. 27 (8), 608–609. DOPA decarboxylase may be stimulated. Complementary [11] Pickens, J.T. (1981) Sedative activity of cannabis in relation to its experiments testing the role of CBD on the biosynthesis of cat- D0-trans-tetrahydrocannabinol and cannabidiol content. Br. J. echolamines would provide us a better understanding of the Pharmacol. 72 (4), 649–656. [12] Russo, E. and Guy, G.W. (2006) A tale of two cannabinoids: the phenomena. therapeutic rationale for combining tetrahydrocannabinol and CBD has been tested lately as a therapeutical agent [14,59– cannabidiol. Med. Hypotheses 66 (2), 234–246. 64]. The present results indicate an additional potential thera- [13] Guimaraes, F.S., Chiaretti, T.M., Graeff, F.G. and Zuardi, A.W. peutic benefit of this compound such as excessive somnolence. (1990) Antianxiety effect of cannabidiol in the elevated plus-maze. Early studies have indicated that sleep may be modulated by Psychopharmacol 100 (4), 558–589. [14] Lastres-Becker, I., Molina-Holgado, F., Ramos, J.A., Mechou- cannabinoids [15–17] or cannabis extracts. In most experi- lam, R. and Fernandez-Ruiz, J. (2005) Cannabinoids provide ments with extracts however the levels of CBD were not mea- neuroprotection against 6-hydroxydopamine toxicity in vivo and sured. Here we describe the waking properties of CBD. It in vitro: relevance to Parkinson’s disease. Neurobiol. Dis. 19 (1– might be considered to treat sleep disorders such as excessive 2), 96–107. somnolence. This common sleep disturbance is defined as trou- [15] Pivik, R.T., Zarcone, V., Dement, W.C. and Hollister, L.E. (1972) D-9-tetrahydrocannabinol and synhexl: effects on human sleep ble falling asleep or staying asleep. It can cause sleepiness or patterns. Clin. Pharmacol. Ther. 13 (3), 426–435. fatigue during the day, may affect mood and result in trouble [16] Feinberg, I., Jones, R., Walker, J.M., Cavness, C. and March, J. focusing on tasks. According to National Sleep Foundation re- (1975) Effects of high dosage D-9-tetrahydrocannabinol on sleep sults from the 2005 Sleep in America poll indicate that 60% of patterns in man. Clin. Pharmacol. Ther. 17 (4), 458–466. [17] Feinberg, I., Jones, R., Walker, J., Cavness, C. and Floyd, T. America’s adults who drive or have a license report that, with- (1976) Effects of marijuana extract and tetrahydrocannabinol on in the past year, they have driven a car or motor vehicle when electroencephalographic sleep patterns. Clin. Pharmacol. Ther. 19 feeling drowsy. We now could consider the use of CBD, the (6), 782–794. non-psychoactive constituent of C. sativa, to treat sonmolence. [18] Monti, J.M. (1977) Hypnotic-like effects of cannabidiol in the rat. In summary, administration of CBD to rats produced wak- Psychopharmacology 55 (3), 263–265. [19] Carlini, E.A. and Cunha, J.M. (1981) Hypnotic and antiepileptic ing in association with activation of hypothalamic and DRD, effects of cannabidiol. J. Clin. Pharmacol. 21 (Suppl. 8–9), 417S– two brain regions implicated in the control of alertness. Extra- 427S. 4344 E. Murillo-Rodrı´guez et al. / FEBS Letters 580 (2006) 4337–4345

[20] Nicholson, A.N., Turner, C., Stone, B.M. and Robson, P.J. wakefulness determined by microinjection of muscimol in freely (2004) Effect of D-9-tetrahydrocannabinol and cannabidiol on moving cats. Brain Res. 479, 225–240. nocturnal sleep and early-morning behavior in young adults. J. [39] McGinty, D.J. and Harper, R.M. (1976) Dorsal raphe neurons: Clin. Psychopharmacol. 24 (3), 305–313. depression of firing during sleep in cats. Brain Res. 101 (3), 569– [21] Tanda, G., Pontieri, F.E. and Di Chiara, G. (1997) Cannabinoid 575. and heroin activation of mesolimbic dopamine transmission by a [40] Trulson, M.E. and Jacobs, B.L. (1979) Raphe unit activity in common mu1 opioid receptor mechanism. Science 276 (5321), freely moving cats: correlation with level of behavioral arousal. 2048–2050. Brain Res. 163 (1), 135–150. [22] Malone, D.T. and Taylor, D.A. (1999) Modulation by fluoxetine [41] Jones, B.E. (2003) Arousal systems. Front. Biosci. 8, s438–s451. of striatal dopamine release following D9-tetrahydrocannabinol: a [42] Russo, E.B., Burnett, A., Hall, B. and Parker, K.K. (2005) microdialysis study in conscious rats. Br. J. Pharmacol. 128 (1), Agonistic properties of cannabidiol at 5-HT1a receptors. Neuro- 21–26. chem. Res. 30 (8), 1037–1043. [23] van der Stelt, M. and Di Marzo, V. (2003) The endocannabinoid [43] McPartland, J.M. and Russo, E.B. (2001) Cannabis and cannabis system in the basal ganglia and in the mesolimbic reward system: extracts: greater than the sum of their parts? J. Cannabis Ther. 1 implications for neurological and psychiatric disorders. Eur. J. (3–4), 103–132. Pharmacol. 480 (1–3), 133–150. [44] Jones, B.E., Bobillier, P., Pin, C. and Jouvet, M. (1973) The effect [24] Bisogno, T., Hanus, L., De Petrocellis, L., Tchilibon, S., Ponde, of lesions of catecholamine-containing neurons upon monoamine D.E., Brandi, I., Moriello, A.S., Davis, J.B., Mechoulam, R. and content of the brain and EEG and behavioral waking in the cat. Di Marzo, V. (2001) Molecular targets for cannabidiol and its Brain Res. 58 (1), 157–177. synthetic analogues: effect on vanilloid VR1 receptors and on the [45] Rye, D.B. and Jankovic, J. (2002) Emerging views of dopamine in cellular uptake and enzymatic hydrolysis of anandamide. Br. J. modulating sleep/wake state from an unlikely source: PD. Pharmacol. 134 (4), 845–852. Neurology 58 (3), 341–346. [25] Fride, E., Feigin, C., Ponde, D.E., Breuer, A., Hanus, L., [46] Paus, S., Brecht, H.M., Koster, J., Seeger, G., Klockgether, T. Arshavsky, N. and Mechoulam, R. (2004) (+)-Cannabidiol and Wullner, U. (2003) Sleep attacks, daytime sleepiness, and analogues which bind cannabinoid receptors but exert peripheral dopamine agonists in Parkinson’s disease. Mov. Disord. 18 (6), activity only. Eur. J. Pharmacol. 506 (2), 179–188. 659–667. [26] Murillo-Rodriguez, E., Sanchez-Alavez, M., Navarro, L., Marti- [47] Isaac, S.O. and Berridge, C.W. (2003) Wake-promoting actions of nez-Gonzalez, D., Drucker-Colin, R. and Prospero-Garcia, O. dopamine D1 and D2 receptor stimulation. J. Pharmacol. Exp. (1998) Anandamide modulates sleep and memory in rats. Brain Ther. 307 (1), 386–394. Res. 812 (1–2), 270–274. [48] Nishino, S., Mao, J., Sampathkumaran, R. and Shelton, J. (1998) [27] Murillo-Rodriguez, E., Cabeza, R., Mendez-Diaz, M., Navarro, Increased dopaminergic transmission mediates the wake-promot- L. and Prospero-Garcia, O. (2001) Anandamide-induced sleep is ing effects of CNS stimulants. Sleep Res. Online 1 (1), 49–61. blocked by SR141716A, a CB1 receptor antagonist and by [49] Kume, K., Kume, S., Park, S.K., Hirsh, J. and Jackson, F.R. U73122, a phospholipase C inhibitor. Neuroreport 12 (10), 2131– (2005) Dopamine is a regulator of arousal in the fruit fly. J. 2136. Neurosci. 25 (32), 7377–7384. [28] Murillo-Rodriguez, E., Blanco-Centurion, C., Sanchez, C., Piom- [50] Lena, I., Parrot, S., Deschaux, O., Muffat-Joly, S., Sauvinet, V., elli, D. and Shiromani, P.J. (2003) Anandamide enhances Renaud, B., Suaud-Chagny, M.F. and Gottesmann, C. (2005) extracellular levels of adenosine and induces sleep: an in vivo Variations in extracellular levels of dopamine, noradrenaline, microdialysis study. Sleep 26 (8), 943–947. glutamate, and aspartate across the sleep–wake cycle in the medial [29] Paxinos, G. and Watson, C. (1986) The Rat Brain in Stereotaxic prefrontal cortex and nucleus accumbens of freely moving rats. J. Coordinates, Academic Press, New York. Neurosci. Res. 81 (6), 891–899. [30] Mechoulam, R., Parker, L.A. and Gallily, R. (2002) Cannabidiol: [51] Lu, J., Jhou, T.C. and Saper, C.B. (2006) Identification of wake- an overview of some pharmacological aspects. J. Clin. Pharmacol. active dopaminergic neurons in the ventral periaqueductal gray 42 (Suppl. 11), 11S–19S. matter. J. Neurosci. 26 (1), 193–202. [31] Verdugo-Diaz, L., Palomero-Rivero, M. and Drucker-Colin, R. [52] Wisor, J.P., Nishino, S., Sora, I., Uhl, G.H., Mignot, E. and (1998) Differentiation of chromaffin cells by extremely low Edgar, D.M. (2001) Dopaminergic role in stimulant-induced frequency magnetic fields changes ratios of catecholamine type wakefulness. J. Neurosci. 21 (5), 1787–1794. messenger. Bioelectrochem. Bioenerg. 46, 297–300. [53] Drysdale, A.J., Ryan, D., Pertwee, R.G. and Platt, B. (2006) [32] Garcia-Garcia, F., Beltran-Parrazal, L., Jimenez-Anguiano, A., Cannabidiol-induced intracellular Ca(2+) elevations in hippo- Vega-Gonzalez, A. and Drucker-Colin, R. (1998) Manipulations campal cells. Neuropharmacology 50 (5), 621–631. during forced wakefulness have differential impact on sleep [54] Malfait, A.M., Gallily, R., Sumariwalla, P.F., Malik, A.S., architecture, EEG power spectrum, and Fos induction. Brain Andreakos, E., Mechoulam, R. and Feldmann, M. (2000) The Res. Bull. 47 (4), 317–324. nonpsychoactive cannabis constituent cannabidiol is an oral anti- [33] Szymusiak, R., Alam, N., Steininger, T.L. and McGinty, D. arthritic therapeutic in murine collagen-induced arthritis. Proc. (1998) Sleep-waking discharge patterns of ventrolateral preoptic/ Natl. Acad. Sci. USA 97 (17), 9561–9566. anterior hypothalamic neurons in rats. Brain Res. 803 (1–2), 178– [55] Parker, L.A., Mechoulam, R. and Schlievert, C. (2002) Canna- 188. bidiol, a non-psychoactive component of cannabis and its [34] Guimaraes, V.M., Zuardi, A.W., Del Bel, E.A. and Guimaraes, synthetic dimethylheptyl homolog suppress nausea in an exper- F.S. (2004) Cannabidiol increases Fos expression in the nucleus imental model with rats. Neuroreport 13 (5), 567–570. accumbens but not in the dorsal striatum. Life Sci. 75 (5), 633– [56] Devane, W.A., Hanus, L., Breuer, A., Pertwee, R.G., Stevenson, 638. L.A., Griffin, G., Gibson, D., Mandelbaum, A., Etinger, A. and [35] Sallanon, M., Kitahawa, K., Buda, C., Puymartin, M., Luppi, Mechoulam, R. (1992) Isolation and structure of a brain P.H. and Jouvet, M. (1986) Effect of electrolytic lesion of constituent that binds to the cannabinoid receptor. Science 258 hypothalamic paraventricular nucleus and its related areas on (5090), 1946–1949. the sleep-waking cycle in the cat. Arch. Ital. Biol. 125, 305– [57] Rodriguez de Fonseca, F., Navarro, M., Gomez, R., Escuredo, 315. L., Nava, F., Fu, J., Murillo-Rodriguez, E., Giuffrida, A., [36] Suntsova, N.V., Dergacheva, O.Y. and Burikov, A.A. (2000) The LoVerme, J., Gaetani, S., Kathuria, S., Gall, C. and Piomelli, role of the posterior hypothalamus in controlling the paradoxical D. (2001) An anorexic lipid mediator regulated by feeding. Nature phase of sleep. Neurosci. Behav. Physiol. 30 (2), 161–167. 414 (6860), 209–212. [37] Suntsova, N.V. and Dergacheva, O.Y. (2003) Dynamics of [58] Fegley, D., Gaetani, S., Duranti, A., Tontini, A., Mor, M., neuron activity in the lateral preoptic area of the hypothalamus Tarzia, G. and Piomelli, D. (2005) Characterization of the fatty during the sleep-waking cycle. Neurosci. Behav. Physiol. 33 (7), acid amide hydrolase inhibitor cyclohexyl carbamic acid 30- 651–658. carbamoyl-biphenyl-3-yl ester (URB597): effects on anandamide [38] Lin, J.S., Sakai, K., Vanni-Mercier, G. and Jouvet, M. (1989) A and oleoylethanolamide deactivation. J. Pharmacol. Exp. Ther. critical role of the posterior hypothalamus in the mechanism of 313 (1), 352–358. E. Murillo-Rodrı´guez et al. / FEBS Letters 580 (2006) 4337–4345 4345

[59] Merritt, J.C., Crawford, W.J., Alexander, P.C., Anduze, A.L. and [62] Hampson, A.J., Grimaldi, M., Axelrod, J. and Wink, D. Gelbart, S.S. (1980) Effect of marihuana on intraocular and blood (1998) Cannabidiol and ()D9-tetrahydrocannabinol are neu- pressure in glaucoma. Ophthalmology 87 (3), 222–228. roprotective antioxidants. Proc. Natl. Acad. Sci. USA 95 (14), [60] Welch, S.P. and Stevens, D.L. (1992) Antinociceptive activity of 8268–8273. intrathecally administered cannabinoids alone, and in combination [63] Mechoulam, R. (1999) Recent advantages in cannabinoid with , in mice. J. Pharmacol. Exp. Ther. 262 (1), 10–18. research. Forsch Komplementarmed Suppl 3, 16–20. [61] Abrahamov, A., Abrahamov, A. and Mechoulam, R. (1995) An [64] Teare, L. and Zajicek, J. (2005) The use of cannabinoids in efficient new cannabinoid antiemetic in pediatric oncology. Life multiple sclerosis. Expert Opin. Investig. Drugs 14 (7), 859– Sci. 56 (23–24), 2097–2102. 869. RESEARCH ARTICLE Endocannabinoid Signaling Regulates Sleep Stability

Matthew J. Pava1, Alexandros Makriyannis2, David M. Lovinger1*

1 Section on Synaptic Pharmacology, Laboratory for Integrative Neuroscience, Division of Intramural Biological and Clinical Research, National Institute on Alcohol Abuse and Alcoholism, National Institutes of Health, Rockville, MD, United States of America, 2 Center for Drug Discovery and Departments of Chemistry and Chemical Biology and Pharmaceutical Sciences, Northeastern University, Boston, Massachusetts, United States of America

* [email protected]

Abstract

The hypnogenic properties of cannabis have been recognized for centuries, but endoge- nous cannabinoid (endocannabinoid) regulation of vigilance states is poorly characterized. We report findings from a series of experiments in mice measuring sleep with polysomno- OPEN ACCESS graphy after various systemic pharmacological manipulations of the endocannabinoid sys- Citation: Pava MJ, Makriyannis A, Lovinger DM tem. Rapid, unbiased scoring of vigilance states was achieved using an automated (2016) Endocannabinoid Signaling Regulates Sleep Stability. PLoS ONE 11(3): e0152473. doi:10.1371/ algorithm that we devised and validated. Increasing endocannabinoid tone with a selective journal.pone.0152473 inhibitor of monoacyglycerol lipase (JZL184) or fatty acid amide hydrolase (AM3506) pro-

Editor: Eric M Mintz, Kent State University, UNITED duced a transient increase in non-rapid eye movement (NREM) sleep due to an augmenta- STATES tion of the length of NREM bouts (NREM stability). Similarly, direct activation of type 1

Received: November 8, 2015 cannabinoid (CB1) receptors with CP47,497 increased NREM stability, but both CP47,497 and JZL184 had a secondary effect that reduced NREM sleep time and stability. This sec- Accepted: March 15, 2016 ondary response to these drugs was similar to the early effect of CB1 blockade with the Published: March 31, 2016 antagonist/inverse agonist AM281, which fragmented NREM sleep. The magnitude of the Copyright: This is an open access article, free of all effects produced by JZL184 and AM281 were dependent on the time of day this drug was copyright, and may be freely reproduced, distributed, administered. While activation of CB1 resulted in only a slight reduction in gamma power, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made CB1 blockade had dramatic effects on broadband power in the EEG, particularly at low fre- available under the Creative Commons CC0 public quencies. However, CB1 blockade did not significantly reduce the rebound in NREM sleep domain dedication. following total sleep deprivation. These results support the hypothesis that endocannabi- Data Availability Statement: All relevant data are noid signaling through CB1 is necessary for NREM stability but it is not necessary for sleep within the paper and its Supporting Information files. homeostasis. Funding: This work was supported by the United States National Institutes of Health. DML and MJP received support from the National Institute on Alcohol Abuse and Alcoholism's Division of Intramural Clinical and Biological Research, award Introduction number ZIA AA000416 (http://www.niaaa.nih.gov/). Since antiquity cannabinoids have been used as a treatment for insomnia [1], and the first AM received support from National Institute on Drug reports in western medical literature regarding the therapeutic utility and physiological effects Abuse Grant DA003801 (http://www.drugabuse.gov/). – The funders had no role in study design, data of cannabis preparations note their hypnogenic properties [2 5]. Additionally, this effect collection and analysis, decision to publish, or appears to be conserved across mammalian species [6–11]. Given the long standing recognition preparation of the manuscript. of cannabinoids as sleep promoting substances, it is surprising that relatively few studies have

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 1/47 Endocannabinoid Signaling Regulates Sleep Stability

Competing Interests: The authors have declared examined the role of the endogenous cannabinoid (endocannabinoid; eCB) system in regulat- that no competing interests exist. ing vigilance states. Cannabinoids produce the majority of their central effects by activating the cannabinoid 1 receptor (CB1), and activation of this G-protein-coupled receptor (GPCR) reduces neurotrans- mitter release at many synapses [12]. CB1 is a central molecular component of the eCB system, an increasingly well characterized, lipid-based neuromodulatory system. The predominant transmitters for the eCB system are N-arachidonyl ethanolamide (anandamide; AEA) and 2-archidonylglycerol (2-AG). These molecules are released during periods of neuronal activity, and their inactivation occurs largely via distinct hydrolytic pathways. AEA is primarily inacti- vated via fatty acid amide hydrolase (FAAH), and 2-AG signaling is terminated by monoacy- glycerol lipase (MAGL). Of the relatively few studies that have been performed, administration of exogenous AEA consistently increases rapid eye movement (REM) sleep and non-REM (NREM) sleep [13–16]. However, conflicting results arise from attempts to increase endoge- nous AEA levels. Some studies indicate that FAAH inhibition promotes wake [17, 18], but other reports show that blocking the AEA membrane transporter facilitates NREM sleep [19, 20]. Additionally, mice with a constitutive knockout of FAAH have increased NREM sleep time and reduced wake [21]. The effects of MAGL inhibition on sleep have not been examined. While cannabinoids have been used by humans for many years to increase sleep, patients in clinical trials for the CB1 antagonist/inverse agonist, rimonabant, commonly reported sleep dis- turbances [22, 23]. In support of a sleep promoting role of eCB signaling, several studies have found fragmented sleep in CB1-null mutant mice [24, 25]. However, studies with constitutive knockout mice are always subject to confounds arising from developmental adaptations, and this has been confirmed for the CB1 knockout mice used in these studies [26, 27]. On the other hand, studies with CB1 antagonists in rodents have had conflicting results with some reporting a weak reduction in NREM sleep [15, 19, 28–30] and others finding no effects on sleep [13, 31, 32]. Of note, all of these studies were performed over short time windows (< 8 Hr recordings), and eCB levels are known to fluctuate over the circadian cycle [33, 34]. Thus, differences in the time of day these experiments were performed could explain some of this discrepancy. As there is a poor consensus regarding the effects of eCB signaling on sleep, we performed a series of experiments comprising over 11,000 Hr of polysomnographic recordings in mice fol- lowing a variety of pharmacological manipulations probing different aspects of the eCB system. To more fully account for the time course of effects, sleep measures were assessed over a 23.5 Hr period following all manipulations. To analyze this large volume of data, we developed and validated a novel automated state-scoring algorithm. In addition to a description of sleep, we also report results from power spectral analyses of electroencephalographic (EEG) recordings. Finally, we directly tested whether eCB signaling is necessary for homeostatic regulation of sleep by blocking CB1 signaling during recovery from total sleep deprivation (TSD). Our find- ings indicate that eCB signaling is both necessary and sufficient to promote long (stable) bouts of NREM sleep, but eCBs are not necessary for sleep homeostasis. These findings constitute a thorough characterization of eCB modulation of vigilance states that should provide a platform for future studies examining the physiological mechanisms of eCB regulation of sleep.

Methods Ethics Statement This research involved the use of vertebrate animals (mice), including survival surgical proce- dures to implant electrodes. All methods were approved by the Institutional Animal Care and Use Committee of the National Institute on Alcohol Abuse and Alcoholism (protocol #: LIN-DL-22) and hewed to guidelines specified in the Guide to the Care and Use of Laboratory

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 2/47 Endocannabinoid Signaling Regulates Sleep Stability

Animals [35]. Survival surgeries were performed under anesthesia, and ketoprofen (5 mg/kg i.p.) analgesic solution was administered immediately after surgery and every 24 hr for the next two days.

Subjects Male C57BL/6J mice (103, 8–10 week olds) were obtained from the Jackson Laboratory (Bar Har- bor, ME), and initially group housed, 2–4 mice per cage. Mice weighed 25–30 g at the beginning of the study, and body weight did not change substantially over the course of experiments. Fol- lowing surgery, subjects were single housed for the remainder of the study. At all times, subjects were provided with ad libitum food and water. The colony and sleep recording environment were maintained on a 12 hr light:dark cycle with the light photoperiod (LP) starting at 06:30 and the dark photoperiod (DP) beginning at 18:30. For the experiment where JZL 184 was adminis- tered prior to the LP, mice were housed in reverse cycle conditions with lights turning on at 18:30 and off at 06:30 for 2 weeks prior to recordings and throughout the recording period. Time of day is expressed throughout this manuscript relative to the light zeitgeber (ZT) with ZT 00:00 coinciding with beginning of LP and ZT 12:00 coinciding with the beginning of the DP. The col- ony and recording environment were maintained at 22.2°C and 50% humidity.

Surgical Implantation of Electrodes Prior to surgery, custom implants were prepared. One end of three single-stranded, Teflon coated stainless steel wires (#791500, A-M Systems, Sequim, WA) was soldered to individual gold-plated sockets (E363/0, PlasticsOne, Roanoke, VA). These three gold sockets and the socket attached to a stainless steel suture pad (E363T/2, PlasticsOne) were arranged in a plastic 6 channel connector (MS363, PlasticsOne) and secured with non-conductive epoxy. During surgery, two of the stainless steel wires emerging from the implant were wrapped, separately, around the frontal electrodes to provide two EEG channels. The ground electrodes were shorted together with the remaining wire. To ensure electrical connectivity with the EEG and ground electrodes a small amount of electrically conductive glue (Bare Paint, Bare Conductive Ltd., London, UK) was applied at the junction between wires and the stainless steel screws. Stereotaxic surgery was performed to implant subjects with EEG/EMG electrodes. EEG elec- trodes consisting of stainless steel screws (Small Parts# AMS90/1P-25, Amazon Supply, Seattle, WA) were implanted supradurally through the skull. Two electrodes were implanted over frontal cortex (B: RC +2.64, ML ± 1.38) and referenced to two, connected ground electrodes implanted over occipital cortex (B: RC—2.5, ML ± 2). The EMG electrode (metalsuturepad,PlasticsOne,Roa- noke, VA) was implanted underneath the nuchal muscle. A head cap was formed with standard, cold-cure dental acrylic, and subjects were allowed to recuperate for two weeks in their home cages.

Sleep Recordings Following recuperation from surgery, subjects were lightly anesthetized with isoflurane and con- nected to a non-motorized commutator (SL6C/SB, Plastics One) via an electrical tether. Subjects were placed into a recording home cage fabricated from a 4 liter, clear polycarbonate bucket (Cambro RFSCW4135, Webstaurant Store, Lancaster, PA). These cages contained standard corn- cob bedding, and food pellets were placed on the cage floor. Access to water was provided via glass liquid diet feeding tubes (#9019, Bio-serve, Frenchtown, NJ) inserted through a hole drilled through the side of each cage. The commutators were secured to a hole in the cage lid thus ensur- ing that mice did not become entangled in their tethers. Five cages were placed inside sound and light attenuating chambers equipped with a fan and white LED light strips (# 10434, General Elec- tric, Fairfield, CT). The lights were on a timer synchronized with the colony lights and the data

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 3/47 Endocannabinoid Signaling Regulates Sleep Stability

acquisition PC clock. Additionally, the inside of the chamber was lined with either copper mesh or coated in conductive paint to shield the inside from electromagnetic interference. Prior to recording, mice were habituated to this environment for 7 days. On recording days, data were collected over a 23.5 Hr period. The time at which recordings were initiated depended on the experimental protocol, but this was generally either before the onset of the LP (ZT 00:00) or before the onset of the DP (ZT 12:00). For the sleep deprivation experiment, data collection was initiated in the middle of the LP (ZT 06:00). Individual cages were removed from the recording chambers during the 30 min between recording sessions, and subjects were weighed and administered an i.p. injection of saline, vehicle, or a drug. After injections, subjects were placed back into their respective cages, and these were returned to the recording cham- bers. In studies with multiple doses of drugs, we explicitly chose to use a schedule of escalating doses over a series of days. Cannabinoids can rapidly induce tolerance, and thus, if a counter- balanced design was implemented, any effect of low-dosage on sleep parameters would likely be prevented by the preceding administration of a high dose. Additionally, compensatory effects were observed following high dose administration of cannabinoids (for example, see results from the recovery day in experiments with JZL184 and AM3506), so a counterbalanced design on successive days could lead to erroneous conclusions that low dosage administration promotes effects opposing those of high-doses. EEG and EMG signals were amplified 1000x (20x HST/16V-G20 headstage followed by 50x wide-band PBX, Plexon, Dallas, TX). The amplified signals were digitized using a National Instruments digitizer (PCI-6071E, National Instruments, Austin, TX) connected to a standard PC computer (Optiplex GX620, Dell Computers, Round Rock, TX) running Recorder v2 (Plexon). Data were visualized as they were being collected using Recorder software’s built-in oscilloscope, and in cases where signals were observed to be of poor quality (e.g. low signal-to- noise ratio, low dynamic range), the subject was not used for analysis of sleep or drug-induced changes in EEG spectral power. This was almost always the result of a faulty EMG electrode. As an additional check to ensure subjects with corrupt (unscorable) signals were not included in analysis, subsequent visual inspection of the scoring results (including raw EEG/EMG sig- nals) for each subject alerted the investigator to any problems of low signal quality that could arise at a point in the experiment when online data collection was not being monitored. All data were sampled at 1 kHz, and a 60 Hz notch filter was applied to eliminate line noise. EEG signals were low-pass filtered online at 120 Hz with a 2 pole Bessel filter, and EMG signals were high pass filtered at 40 Hz with a 4 pole Bessel filter. Data were saved for offline analysis.

Sleep Deprivation The custom sleep deprivation apparatus (S1 Fig, panel A) used in this report was constructed from a 10 in long clear acrylic tube with an internal diameter of 5 in that was suspended slightly above an epoxy-sealed ABS disc that served as the chamber floor. The disc and other structural components of the apparatus were custom designed in CAD software (Sketchup 8, Trimble Navigation Ltd, Sunnyvale, CA) and constructed from 3D printed ABS plastic (3D XL printer, Airwolf3D, Costa Mesa, CA). The cylinder had an acrylic divider that extended the length of its radius from the interior wall directly into the center of the chamber (S1 Fig, panel B). This divider ensured the animal would move when the floor of the device rotated. The chambers contained standard corncob bedding, and a sufficient amount of standard mouse chow was placed on the floor to provide ad libitum access to food. A glass liquid diet feeding tube pro- vided ad libitum access to water. The bottom of ABS disc was attached to a 360 degree servo motor (DF15RSMG, DFRobot, Shanghai, China), and the motors of five chambers were con- trolled by an Arduino UNO board (Adafruit Industries, New York, NY) that received

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 4/47 Endocannabinoid Signaling Regulates Sleep Stability

commands via a serial connection with a laptop running MATLAB (S1 Fig, panel C). The sig- nal from the Arduino board and the servos were calibrated to rotate at approximately 15 rpm. Rotation was engaged for a random time interval between 10–15 sec and was turned off for a random interval between 5–10 sec. Additionally, the rotational direction of the disc was ran- domly alternated to prevent habituation to movement in one direction. In preliminary studies, we found this intermittent schedule to be effective in producing near total sleep deprivation (< 1% of time spent in NREM sleep) for up to 6 Hr in most subjects. Importantly, the devices were constructed so that a commutator (SL6C/SB, Plastics One) could be connected to the chamber lid. This allowed continuous recording of the EEG and EMG signals while the subject was housed in the apparatus. Sleep deprivation experiments consisted of three phases: baseline, deprivation, and recovery. Polysomnographic measures of sleep were obtained across all phases of the experiment. Fol- lowing habituation to the standard recording environment, 48 Hr baseline recordings were obtained from all subjects following an i.p. injection of vehicle (1:1:18 mixture DMSO: Crema- phor: 0.9% Saline) at ZT 06:00. After the baseline recordings, subjects were transferred into the sleep deprivation devices (S1 Fig). Importantly, their original recording cages (including bed- ding, food, and water bottles) were retained and labelled according to subject. TSD via forced locomotion was initiated at the onset of the LP (ZT 00:00), 18 Hr after the subjects were placed into the deprivation device. The TSD protocol continued for 6 Hr, and afterwards subjects were immediately removed from the deprivation chambers, weighed, and received either an i. p. injection of the vehicle solution (control) or 5 mg/kg AM281. Subjects were then returned to their original recording cage for 48 Hr (recovery phase). TSD was defined as spending less than 1% of total time (< 3.6 minutes) in NREM sleep, and based on this criterion, we disqualified 5 out of 14 mice in the AM281 group and 3 out of 14 mice in the vehicle group.

Vigilance State Scoring To obtain an unbiased estimate of sleep-wake states, we devised an automated algorithm to score polysomnographic data as either wake, NREM, or REM sleep (Fig 1A and 1B). Impor- tantly, this software arrives at a deterministic score of a 24 Hr single-subject recording in less than 5 min, assigning scores to 2 sec epochs, and it performs as well as trained human scorers (Fig 1C and S3 Fig). Calculation of the State-Space. The first step, deriving the state-space (Fig 1A), was heavily influenced by the state-space methodology reported by Gervasoni et al [36], but in addition to electrographic signals from the brain, we also incorporated EMG activity to con- form with standard polysomnography techniques used in mice [37]. However, our methodol- ogy differed somewhat from other state-space based approaches. Because we were recording from two EEG channels, we first compressed this data by taking the first principle component (PC) of the raw data. By only performing analysis on the 1st PC of the two frontal EEG signals, computational overhead is reduced. Next, power spectra were obtained for the EEG and EMG waveforms using a 4 sec sliding window FFT with a 2 sec step. This was implemented using the spectrogram() function that is part of the signal processing toolbox in MATLAB (The Math- works Inc, Natick, MA). This provided frequency domain data in 2 sec epochs with roughly 0.25 Hz bin resolution. Two power spectral ratios (R1 & R2) were calculated from the EEG data for each epoch:

P20 P10 Hz Pðf Þ Hz Pðf Þ 1 1 Pj¼0:5 Hz j 2 2 P j¼5 Hz j Ratio ¼½R i¼ 100 Hz Ratio ¼½R i¼ 4 Hz j¼0:5 Hz PðfjÞ j¼0:5 Hz PðfjÞ

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 5/47 Endocannabinoid Signaling Regulates Sleep Stability

Fig 1. Design and Validation of Fully Automated Vigilance State Scoring Algorithm. A, Schematic diagram of data processing during acquisition (blue shaded region) and during offline calculation of 3-dimensional state-space coordinates (green shaded region). The grey box at the end of the flow chart highlights the three coordinates that define the state space. At this stage, no state-assignment has been made. Each point in the final state-space represents a 2 sec epoch. B, Schematic illustrating the three-step, automated process for classifying points in the state-space (pink shaded region) into either wake, rapid eye movement (REM), or non-rapid eye movement (NREM) sleep. Additionally, points that are ambiguously positioned on the boundary between clusters can be defined as unclassified. Step 1: parametric classification establishes regions of the state-space consistent with the three vigilance states based on hard cutoff criteria determined from the distribution of points within the state-space. Step 2a: from the classification performed in step 1, 99% confidence intervals (CIs) are constructed for each state using a product kernel estimator with a Gaussian kernel function and Scott’s Rule for bandwidth determination. Step 2b: The state-space is reclassified using a simple inclusion rule with the 99% CIs constructed in step 2a. Step 3: A transitional classifier is used to incorporate most points that were outside the 99% CIs into a state-classification. All strings of unclassified points that are bounded on either side by epochs of the same state are incorporated into that state classification (e.g. wake–unclassified–unclassified–wake becomes wake–wake–wake–wake). C, Validation results comparing the percent agreement between three trained human scorers (inter-rater reliability) with percent agreement between each human and the computer assigned scores. Bars represent mean±SEM. Abbreviations: CI–confidence interval, EEG–electroencephalogram, EMG– electromyogram (A) or state-space coordinate derived from electromyogram signal (B), FFT–Fast Fourier Transform, HP–high pass, LP–low pass, PC– principle component, PCA–principle component analysis, R1 –ratio 1, R2 –ratio 2. doi:10.1371/journal.pone.0152473.g001

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 6/47 Endocannabinoid Signaling Regulates Sleep Stability

Ratio 1 (R1) provides good separation between NREM and wake epochs [36, 38]. However, separating REM from the other two clusters is challenging due to the sparse nature of REM sleep and the similarity of the power spectra between REM and wake in mice. However, theta rhythms (5–8 Hz) are prominent during REM sleep, and this characteristic has been used as a means to separate REM epochs into a distinct, if somewhat diffuse cluster [38–40]. Thus, we defined Ratio 2 (R2) to help pull epochs with high theta power away from the major clusters representing NREM and wake. In addition to prominent theta, REM sleep is also distinguished from wake by significantly reduced muscle tonus which can be measured with EMG. There- fore, we incorporated this criterion into the state-space by including the RMS value of the power spectra of the EMG waveform. In summary, three criteria were used to separate epochs of polysomnographic data into a 3D state-space: (1) Prominence of the frequencies between 0.5 and 20 Hz relative to the entire power spectra (0.5–100 Hz), (2) Prominence of theta power relative to delta (0.5-4Hz), (3) RMS of the power spectra of the EMG waveform. To provide better cluster separation, the state-space coordinates are smoothed with a 10 sec Hann function and log transformed. To standardize the range of axes occupied by the state- space the log transformed data were median centered and normalized to the max absolute value on each axis. This bounds the state-space between -1 and +1 across all axes for all sub- jects. The first graph in Fig 1B provides an example of the state-space at this stage of algorithm. Classification of Points in State-Space as Vigilance States. Once the state-space is com- puted, three distinct clusters can be observed, and the second step of the classification software (Fig 1B) serves to automatically define these clusters. This process occurs over three sequential steps, where a rough description of clusters based on hard cutoffs is used to establish statistical boundaries (confidence intervals) that reclassify the data using an inclusion/exclusion test. In the final step, the remaining epochs that are not assigned to wake, NREM, or REM states are restricted to epochs that are ambiguous to classify because they occur during transitions between vigilance states. First, a rough estimate of cluster boundaries is obtained by establishing threshold values based on the distribution of points along each axis (Fig 1B, step 1). Specifically, a univariate kernel density estimate is performed independently for EMG and R1. The distribution along R1 is bimodal, so the threshold separating wake and REM from NREM along the R1 axis is defined as the local minimum between these modes. The distribution of EMG values is not consistently bimodal, so NREM epochs are constrained by the third quartile along this axis. Similarly, wake was defined as epochs on the opposite side of the R1 distribution threshold with EMG values greater than the median. REM was defined as values in the same mode of R1 as wake with EMG values within the first tercile. Additionally, REM was constrained to only those epochs above the median value on the R2 axis. The purpose of the these hard-cut criteria is to seed clusters based on the consistent topology of the state-space. The next two steps of the classification process substantially refine this initial classification. The values for thresholds were visually determined by the experimenter after trial and error with many data sets as rea- sonable thresholds to capture the majority of each cluster across the overwhelming majority of datasets tested. Importantly, not all points are classified in this step (see Fig 1B, step 1), and there is a buffer of unclassified points left between wake and REM that is subsequently incorpo- rated into these clusters after completing steps 2 & 3 of the classification process. After inspect- ing the classified state-space for each data file, there were some rare instances where it was clear that the REM cluster was not adequately defined by this step of the algorithm, and in these instances, a custom software routine allowed for the manual selection of the REM cluster. Based upon this initial classification, confidence intervals were calculated for each cluster by ^ estimating the probability density function (PDF) for each cluster separately (f State) and the

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 7/47 Endocannabinoid Signaling Regulates Sleep Stability

^ whole state-space (f total), using kernel estimation with a Gaussian kernel: () ! Xn Y3 1 2 ^ 1 xj Xij t =2 f ðxÞ¼ðnh1h2h3Þ K ; KðtÞ¼pffiffiffiffiffiffi e subset 2p i¼1 j¼1 hj

The smoothing parameter was determined based on the dataset using Scott’s Rule:

^ s^ 1=7 hi ¼ in

Because some points were not assigned a state in the first classification step, there was a sep- arate PDF calculated for these points in addition to the estimates for wake, NREM, and REM. After all kernel estimates had been obtained, they were scaled so the maximum value of each ^ component PDF was equal to the corresponding grid location in f total. In this way, the PDF of the entire state-space was decomposed into component densities representing the different states (Fig 1B, step 2a). ^ ^ ^ ^ ^ f total ¼ f wake þ f NREM þ f REM þ f Unassigned

To determine the probability that a given point belonged to a specified cluster, the compo- nent PDFs were subtracted from one another and normalized to the absolute value of the resulting maxima: ^ ^ ^ ^ x f A f B f C f D PðAj Þ¼ ^ ^ ^ ^ jmaxðf A f B f C f DÞj

Where A, B, C, and D represent different states (Wake, NREM, REM, and Unassigned), and x is a three dimensional feature vector for a specified epoch of the state-space. This subtraction and normalization step was performed for each component density yielding four probability matrices. The subtraction step was important to delineate clean borders between states. At this stage of processing, points in the state-space were reclassified using the probability matrices defined above. To accomplish this, each epoch of the state-space was indexed into the four probability matrices, to determine the probability that it belonged to each state. The epoch was assigned to the state with the highest probability if it fell within confidence intervals speci- fied a priori. We established 99.9% confidence intervals for all states. Points that fell outside of these confidence intervals were assigned to the unclassified cluster, and similarly, points that had equivalent probability of belonging to two or more clusters were assigned to the unclassi- fied cluster (Fig 1B, step 2b). As can be seen in the results following classification with 99.9% confidence intervals (Fig 1B, step 2b), unclassified epochs comprised points on the periphery of clusters and transitional epochs between clusters. To further refine the state assignment, a final classification step was performed using a transitional classifier (Fig 1B, step 3). The point of this last step was to reduce unclassified epochs to only those epochs representing transitions between states where state scoring is inherently ambiguous. Consequently, this classification step assigned all unclas- sified epochs bounded by an epoch of the same state, while unclassified epochs bounded by dif- ferent states would remain unclassified. Thus, the sequence [wake, unclassified, unclassified, wake] would become [wake, wake, wake, wake], while [wake, unclassified, unclassified, NREM] would remain the same. As shown in the last graph of Fig 1B, the result of this classification step was to eliminate the penumbra of unclassified epochs surrounding the clusters, while leav- ing the unclassified epochs between cluster boundaries unchanged.

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 8/47 Endocannabinoid Signaling Regulates Sleep Stability

Most of this analysis was coded in MATLAB and C/C++ (MEX file libraries). However, a parallelized kernel density estimation method, GPUML, was implemented in CUDA/C (NVI- DIA Corp, Santa Clara, CA) to speed computation of non-parametric density estimation [41]. Throughout the algorithm, operations were parallelized where possible, and this was achieved via explicit coding of parfor and spmd loops using MATLAB’s Parallel Computing Toolbox. Importantly, for each data set (all subjects days of experiment), the scoring results were visu- ally examined to ensure there were no obvious defects in scoring. The prevalence of each state was calculated as the percent of time spent in that state. Addi- tionally, the number and duration of bouts of NREM and REM were calculated (sleep architec- ture) with one bout defined as a consecutive series of epochs in the same state.

Drugs CP47,497 (CP47), AM281, JZL184 (JZL), and URB597 (URB) were all obtained from Tocris Bioscience (Bristol, UK). These compounds are highly lipophilic and only sparingly soluble in aqueous solution. Therefore, these drugs were dissolved in a vehicle solution consisting of a 1:1:18 mixture of DMSO, Cremaphor, and normal saline. Administration of this vehicle had no effect on either sleep or EEG power spectra (S2 Fig). AM3506 was synthesized in the labora- tory of Dr. Alexandros Makriyannis (Northeastern University), and was prepared in a 1:1:8 vehicle of DMSO:Cremaphor:Saline because it tends to precipitate when prepared in the stan- dard vehicle solution. For all experiments, 24 Hr recordings of polysomnographic indices fol- lowing administration of the appropriate vehicle solution were used as a within-subject baseline for comparison. All drug and vehicle solutions were administered via i.p. injections given at a volume of 0.02 mL per gram body weight. Drug were prepared fresh on the day of the experiment.

Statistics In all experiments, time of day was included as a factor, but it was necessary to take into account the effect of photoperiod as well. Therefore, all analyses of time course data utilized a hierarchical linear mixed model (HLM) approach. For most experiments, a model with three repeated, fixed factors was implemented. Specifically, the model tested for the interaction between drug treatment and time of day nested within photoperiod. Because we followed sub- jects across different treatment conditions, all analyses contained repeated measures, and post- hoc comparisons were performed within-subjects. For the sleep deprivation experiment, there were two experimental groups (AM281 treated and vehicle control). For simplicity, only data from the first day following baseline vehicle injection (baseline day 1) and the first day follow- ing sleep deprivation (recovery day 1) were statistically compared. In this case, the model design examined the between groups interaction of treatment group (vehicle vs AM281) with time of day nested within photoperiod (light vs dark) nested within experimental phase (base- line vs recovery), where time of day, photoperiod, and experimental phase were all repeated measures. Analysis with HLM was performed in SPSS (IBM, Bethesda, MD). A Bonferroni cor- rection was applied to all pair-wise comparisons of the model-derived estimated marginal means, and all reported P-values reflect this correction. For all analyses α = 0.05.

Results Validation of Unsupervised Sleep Staging Algorithm To validate our method of scoring, the computer-derived vigilance state classification results from 6 datasets were compared against scoring results from three trained humans. Because

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 9/47 Endocannabinoid Signaling Regulates Sleep Stability

manual scoring by humans will always be sensitive to issues of subjectivity and scorer vigilance, an appropriate validation of automated methods should take into account how the computer- derived score compares to the inter-rater reliability of manual scoring. Consequently, the per- cent agreement between scores obtained from the computer algorithm and manual sleep stag- ing were compared to the percent agreement between the manually-derived scores (inter-rater reliability; Fig 1C). There was no interaction between scorer (human vs. computer) and data file (repeated measure; 2-way ANOVA, F(5, 20) = 1.05, p = 0.42), and there was no effect of scorer (F(1, 4) = 1.01, p = 0.37). However, there was an effect of data file (F(5, 20) = 20.76, p < 0.001), because data file 4 was intentionally included as it had a noisy EMG signal. Com- pared to the other files that were scored, there was a marked reduction in the inter-rater reli- ability between humans and between human vs computer derived scores. Comparisons of scoring reliability for each vigilance state also found no difference between humans and the computer (S3 Fig). Consequently, we conclude that this algorithm performs comparably to manual sleep staging. Fig 2 shows example scoring results with raw data traces and power spectra including state transitions. One important feature of this vigilance state-scoring program is the necessary inclusion of unclassified/transitional epochs that cannot be assigned to specific states with any rigor (note the black points between clusters in Fig 2A). This derives naturally from the fact that state clusters are not cleanly segregated in the state-space, which is consistent with the intuitive notion that state-transitions are not instantaneous (i.e. falling asleep or waking up takes some time as cortical ensembles synchronize or desynchronize, respectively). Thus, the algorithm conservatively estimates vigilance states by only assigning a score when an epoch registers within some statistical bounds of certainty.

Direct Activation of CB1 Receptors Facilitates NREM Sleep To determine how activation of CB1 affects sleep, the full CB1 agonist, CP47, was administered just prior to the DP. Consistent with reports that CB1 activation reduces locomotor activity, phasic muscle movements in the EMG were reduced after injection of CP47, and the amount of high voltage, low frequency activity in the EEG was increased (Fig 3). In this experiment, a 0.1 (low) and a 1.0 (high) mg/kg dose of CP47 were administered on subsequent recording days following a baseline day where vehicle was injected (Fig 4A). We assessed the percent time spent in NREM sleep (Fig 4B) and found a significant overall interaction (treatment x time of day within photoperiod, F(18, 142.63) = 9.804, p < 0.001), secondary interaction (treat- ment x photoperiod, F(2, 96.81) = 26.63, p < 0.001), and a main effect of photoperiod (F(1, 116.62) = 284.59, p < 0.001). High dose CP47 had biphasic effects on sleep time, inducing sig- nificantly more NREM during the DP (t(85.57) = 5.71, p < 0.001) and reducing NREM during the LP (t(85.57) = -6.046, p = 0.006). NREM sleep time was increased over the first 6 Hr of the DP (low dose, ZT12-15: t(191.94) = 2.89, p = 0.009; high dose, ZT12-18: t(191.94) 6.21, p < 0.001), and high dose CP47 significantly reduced NREM during the first 3 Hr of the LP (ZT00-03: t(191.94) = -2.54, p = 0.024). Thus, the synthetic cannabinoid CP47 biphasically modulates NREM sleep time. We next examined effects on NREM architecture. For NREM bout duration, there was an overall interaction (treatment x time of day within photoperiod, F(18, 143.87) = 3.854, p < 0.001), a secondary interaction (treatment x photoperiod, F(2, 78.01) = 4.85, p = 0.010), and a main effect of photoperiod (F(1, 113.84) = 26.537, p < 0.001). CP47 significantly increased NREM bout duration during the second quarter of the DP (ZT15-18: t(157.46) = 3.49, p = 0.001). In contrast, NREM bout duration was reduced across the LP (t(62.72) = 3.23, p = 0.032), specifically during the first quarter of the LP (ZT00-03: t(157.46) = 3.16, p = 0.004).

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 10 / 47 Endocannabinoid Signaling Regulates Sleep Stability

Fig 2. Example Results of State-Scoring. A, Example of scored state-space with color-coded state clusters. B, Example of power spectra derived from the scored epochs. Power spectra are color-coded according to the state they were derived from. Solid lines indicate borders of 95% confidence interval of power spectra for all epochs of associated state across the day. C, Distribution of time spent in each classification criteria over the day. D, Pattern of sleep-wake states over the day shown as percent time of 3 Hr bins. Grey background denotes the dark photoperiod. E, Aligned time-frequency power spectrum, raw EEG, raw EMG, and color-coded hypnogram for a single recording day. Time of day denoted as zeitgeber time underneath the hypnogram. F, Expanded view of hashed yellow box in panel E. Small yellow hashed boxes highlight times with state transitions and correspond to subpanels F1 –F3. F1, Wake to NREM transition. F2, NREM to wake transition. F3, Transition from NREM to REM and transition from REM to wake. For A-D and all hypnograms shown in E–F, wake is indicated in red, NREM is indicated in blue, and REM is indicated in green. For all periodograms shown in E and F, absolute power specified in the heat map is given by the colorbar between panel F and subpanel F3. doi:10.1371/journal.pone.0152473.g002

The number of NREM bouts was also affected by an overall interaction (F(18, 143.48) = 1.96, p = 0.016) and a main effect of photoperiod (F(1, 113.30) = 17.81, p < 0.001). However, the only significant difference between treatment conditions occurred during the first quarter of the LP when high dose CP47 increased the number of NREM bouts (t(157.72) = 3.74, p < 0.001). Treatment with CP47 did not affect REM sleep (Fig 4C). To confirm that CP47’s effects on sleep were mediated through the CB1 receptor, a separate cohort of subjects was administered the vehicle solution followed 24 Hr later by an injection containing a mixture of CP47 (1 mg/kg) and the selective CB1 antagonist, AM281 (5 mg/kg;

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 11 / 47 Endocannabinoid Signaling Regulates Sleep Stability

Fig 3. Example EEG/EMG Traces on Different Time Scales Following Vehicle or CP47,497 Administration. EEG and EMG traces are from the same subject at the same stage of the circadian cycle after administration of either vehicle (A, A’, and left column of C) or 1 mg/kg CP47,497 (B, B’, and right column of C). Panels A and B show a 2 Hr 15 min window from ZT 14:00–16:15, roughly 2 Hr after drug administration, coinciding with peak effects observed on sleep. Panels A’ and B’ show a 15 min long segment expanded from the region in A and B highlighted by the dashed orange box. Panel C shows representative 18 sec long data segments corresponding to NREM and Wake obtained following vehicle and CP47 administration. These data segments were taken from the segments shown in A and B. The color-coded hypnogram shown at the bottom of A, B, A’, and B’ represents consecutive 2 sec epochs shown as wake (red), NREM (blue), unclassified (grey). No REM occurred during this period. Black traces depict EEG, red traces depict EMG. A and B are identically scaled. A’ and B’ are identically scaled. All traces in C are identically scaled. doi:10.1371/journal.pone.0152473.g003

CP47+AM281; Fig 4D). As an internal positive control, the experiment was continued for a third day when CP47 (1 mg/kg) was administered alone. There was an overall interaction for NREM sleep time (Fig 4E; treatment x time of day within photoperiod, F(18, 198.22) = 11.31, p < 0.001), a secondary interaction (treatment x photoperiod, F(2, 133.28) = 43.47, p < 0.001), and a main effect of photoperiod (F(1, 160.97) = 364.21, p < 0.001). While CP47 alone increased sleep time during the DP (ZT12-15 & ZT15-18: t(263.34) 6.43, p < 0.001) and reduced it during the LP (ZT00-03, 03–06, and 06–09: t(263.34) -3.12, p 0.006), co-admin- istering AM281 significantly attenuated the CP47 effects during the DP (AM281+CP47 vs CP47, ZT12-15, 15–18, and 18–21: t(263.34) -2.94, p 0.011) and reversed them completely during the LP (ZT00-03 and 03–06: t(263.34) 2.57, p 0.033). Compared to vehicle, CP47 +AM281 had mixed effects on NREM in the DP (ZT12-15 (increased): t(263.34) = 4.14, p < 0.001; ZT18-21(decreased): t(263.34) = -2.79, p = 0.017), but there were no differences

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 12 / 47 Endocannabinoid Signaling Regulates Sleep Stability

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 13 / 47 Endocannabinoid Signaling Regulates Sleep Stability

Fig 4. Direct Activation of CB1 with the Full Agonist, CP47,497, Has Biphasic Effects on NREM Sleep that are Mediated by the CB1 Receptor. A, Diagram of experimental protocol for recording sleep after administration of the CB1 agonist, CP47,497. All injections given at the onset of the dark photoperiod (ZT 12:00). B, Quantification of NREM sleep time and architecture (N = 9). C, Quantification of REM sleep time and architecture. D, Diagram of experimental protocol for recording sleep after co-administering CP47 with AM281. E, Quantification of NREM sleep in experiments with co-administration of CP47 and AM281. In all graphs, the grey shaded region denotes the dark photoperiod. Symbols represent mean±SEM for 3 Hr time bins. Asterisks denote significant difference from vehicle baseline. doi:10.1371/journal.pone.0152473.g004

during the LP. These results replicated our previous findings on CP47’s biphasic effect on sleep, and co-administration of AM281 blocked this effect suggesting that CP47’s effects on NREM sleep are mediated through CB1. The CP47-induced changes in sleep architecture were similarly blunted by co-administra- tion of AM281. For NREM bout duration, there was a significant overall interaction (treatment x time of day within photoperiod, F(18, 202.64) = 5.487, p < 0.001), secondary interaction (treatment x photoperiod, F(2, 103.25) = 13.06, p < 0.001), and main effects of both drug treat- ment (F(2, 61.707) = 4.376, p = 0.017) and photoperiod (F(1, 169.28) = 45.05, p < 0.001). CP47 produced a large decrease in bout duration across all time points of the LP (CP47 vs vehicle, ZT00-12: t(182.30) -2.69, p 0.025) that was blocked by co-administration of AM281 (CP47+AM281 vs CP47, ZT00-09:: t(184.77) 3.30, p 0.003). CP47+AM281 did not change NREM bout duration relative to vehicle. The number of NREM bouts was affected by a signifi- cant overall interaction (treatment x time of day within photoperiod, F(18, 192.19) = 5.20, p < 0.001) and main effects of both photoperiod (F(1, 155.99) = 11.18, p = 0.001) and drug treatment (F(2, 54.76) = 4.79, p = 0.012). Across the entire recording day, CP47 produced an increase in the number of NREM bouts (t(48.65) = 2.98, p = 0.013), but the only specific time point with significant differences between treatments was the first quarter of the DP where both CP47 (t(176.71) = 4.58, p < 0.001) and CP47+AM281 (t(178.81) = -4.57, p < 0.001) pro- duced a significant increase in NREM bouts. As discussed later, AM281 alone increases the number NREM bouts during the first quarter of the DP, so the increased number of NREM bouts is confounded by AM281’s effect. Thus, CP47’s effects on sleep architecture are largely mediated through the CB1 receptor.

Inhibition of Monoacylglycerol Lipase Stabilizes NREM and Suppresses REM Sleep Measurements. Considering that activation of CB1 receptors with exogenous ligands can facilitate sleep, we next sought to test the hypothesis that eCBs could similarly promote NREM sleep. Increasing endogenous 2-AG tone with JZL, a selective MAGL inhibitor, reduced phasic EMG activity and increased the amount of low-frequency high-voltage EEG activity char- acteristic of NREM sleep (Fig 5). To quantify vigilance states after pharmacologically increasing 2-AG levels, subjects were sequentially given a 1.6 (low), 8.0 (moderate), and 16.0 (high) mg/kg doses of JZL, after which an additional 24 Hr recording with no injection (recovery) was obtained (Fig 6A). Within-subject comparisons were made using sleep measures obtained during a 24 Hr baseline recording that followed a vehicle injection. Several reports have suggested that endocan- nabinoid levels fluctuate across the circadian cycle [33, 34, 42], but it is unclear how or if this may be related to sleep. Therefore, two experiments were performed with JZL in separate groups of mice. In one, JZL was administered before the DP, when mice are most active (Fig 6B), and in the other, JZL was administered prior to the LP (Fig 6C). When given just before the onset of the DP (between ZT 11:30 and 12:00), JZL administra- tion had obvious effects on NREM sleep that mirrored those seen with CP47 (Fig 6B; top row).

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 14 / 47 Endocannabinoid Signaling Regulates Sleep Stability

Fig 5. Example EEG/EMG Traces on Different Time Scales Following Vehicle or JZL184 Administration. EEG and EMG traces are from the same subject at the same stage of the circadian cycle after administration of either vehicle (A, A’, and left column of C) or 16 mg/kg JZL184 (B, B’, and right column of C). Data are from experiment with JZL administration before the LP. Panels A and B show a 2 Hr 15 min window from ZT 02:00–04:15, roughly 2 Hr after drug administration, coinciding with peak effects observed on sleep. Panels A’ and B’ show a 15 min long segment expanded from the region in A and B highlighted by the dashed orange box. Panel C shows representative 18 sec long data segments corresponding to NREM and wake obtained following vehicle and JZL administration. These data segments were taken from the segments shown in A and B. The color-coded hypnogram shown at the bottom of A, B, A’, and B’ represents consecutive 2 sec epochs shown as wake (red), NREM (blue), REM (green), and unclassified (grey). Note the loss of REM sleep following JZL administration. Black traces depict EEG, red traces depict EMG. A and B are identically scaled. A’ and B’ are identically scaled. All traces in C are identically scaled. doi:10.1371/journal.pone.0152473.g005

This was evidenced by an overall interaction (treatment x time of day within photoperiod, F(30, 269.56) = 12.29, p < 0.001), secondary interaction (treatment x photoperiod, F(4, 198.66) = 48.31, p < 0.001), and a main effect of photoperiod (F(1, 234.92) = 596.81, p < 0.001). JZL had biphasic effects on NREM with increased sleep during the DP (moderate: t(172.80) = 4.49, p < 0.001; high: t(175.26) = 6.71, p < 0.001) and a suppression of NREM dur- ing the LP (moderate: t(172.80) = -3.72, p = 0.001; high: t(172.80) = -4.62, p < 0.001). Specifi- cally, JZL increased NREM sleep during the middle of the DP (moderate, ZT15-18: t(357.53) = 4.25, p < 0.001; high, ZT15-21: t(357.53) 4.38, p < 0.001) and reduced it during the LP (moderate, ZT00-03: t(357.53) = -3.17, p = 0.007; high, ZT06-09: t(357.53) = -2.93, p = 0.014). In contrast, on the recovery day NREM was reduced during the DP (recovery vs vehicle, t (172.80) = -3.66, p = 0.001) and increased during the LP (t(172.8) = 2.55, p = 0.047), specifically

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 15 / 47 Endocannabinoid Signaling Regulates Sleep Stability

Fig 6. Differential Effects of Increasing 2-AG Tone on NREM and REM Sleep Based on Circadian Timing of Drug Administration. A, Diagram of experimental protocol for recording sleep after administration of the MAGL inhibitor, JZL184. B, Quantification of NREM (top row) and REM (bottom row) sleep time and architecture for experiment where JZL was administered prior to the DP (N = 10). C, Quantification of NREM (top row) and REM (bottom row) sleep time and architecture for experiment where JZL was administered prior to the LP (N = 8). In all graphs, the grey shaded region denotes the dark photoperiod. Symbols represent mean±SEM for 3 Hr time bins. Asterisks denote significant difference from vehicle baseline. doi:10.1371/journal.pone.0152473.g006

during the last quarter of the LP (ZT09-12: t(357.53) = 2.64, p = 0.035). Thus, inhibition of MAGL has biphasic effects on NREM sleep, initially increasing the time in NREM, followed by a decrease in NREM that extends into the recovery day. JZL also produced alterations in NREM architecture similar to CP47. For NREM bout dura- tion, there was an overall interaction (treatment x time of day within photoperiod, F(30, 272.08) = 3.55, p < 0.001), a secondary interaction (treatment x photoperiod, F(4, 153.85) = 20.92, p < 0.001), and main effects of both treatment (F(4, 93.89) = 3.60, p = 0.009) and photo- period (F(1, 230.16) = 4.065, p = 0.045). High dose JZL increased NREM bout duration across the DP (t(115.00) = 2.83, p = 0.022), and both moderate and high doses reduced it across the LP (t(115.00) -2.58, p 0.044). Specifically, high dose JZL increased bout duration during the second quarter of the DP (ZT15-18: t(286.90) = 4.29, p < 0.001) and reduced it across most of the LP (ZT00-09: t(286.90) -2.75, p 0.026). On the recovery day, NREM bout duration was increased across most of the LP (ZT03-12: t(287.54) 2.56, p 0.044). For the number of NREM bouts, there was also an overall interaction (treatment x time of day within

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 16 / 47 Endocannabinoid Signaling Regulates Sleep Stability

photoperiod, F(30, 286.26) = 3.19, p < 0.001), a secondary interaction (treatment x photope- riod, F(4, 153.85) = 3.23, p = 0.014), and main effects of both drug treatment (F(4, 75.96) = 6.15, p < 0.001) and photoperiod (F(1, 272.44) = 99.44, p < 0.001). High dose JZL increased NREM bouts across the LP (t(95.33) = 3.57, p = 0.002) with pair-wise differences at two time points (ZT03-06 & ZT09-12: t(218.71) 2.77, p 0.024). Thus, the effects of JZL treatment on NREM sleep were closely mirrored by effects on NREM bout duration, suggesting that MAGL inhibition-induced changes in sleep are due to modulation of NREM stability. In contrast, administration of JZL before the DP produced only a slight reduction in REM sleep parameters (Fig 6B; bottom row). There was no effect of JZL on REM sleep time. For the duration of REM bouts, there was a nested interaction (time of day within photoperiod, F(6, 238.62) = 10.81, p < 0.001), and main effects of treatment (F(4, 82.54) = 7.01, p < 0.001) and photoperiod (F(1,238.51) = 34.78, p < 0.001). 16.0 mg/kg JZL reduced REM bout duration dur- ing across the DP (t(110.53) = -2.56, p = 0.047), specifically during the third quarter of the DP (ZT18-21: t(235.00) = -2.80, p = 0.022), and REM bout duration was increased across the LP on the recovery day (t(99.54) = 2.77, p = 0.027), specifically during the second quarter of the LP (ZT03-06: t(237.32) = 2.71, p = 0.022). For the number of REM bouts, there was a nested interaction (time of day within photoperiod, (F(6, 268.06) = 14.44, p < 0.001) and main effects of treatment (F(4, 81.95) = 3.17, p = 0.018) and photoperiod (F(1,254.72) = 55.42, p < 0.001). The high dose of JZL increased the number of REM bouts during the last 3 Hr of the LP (ZT09-12: t(240.24) = 3.72, p = 0.001). When JZL was administered before the LP, NREM sleep was augmented (Fig 6C; top row), but the magnitude of this effect (deviation from vehicle baseline) was not as great as when the drug was given prior to the DP. The effect was also not biphasic within a circadian cycle. For the percent of time spent in NREM sleep, there was a secondary interaction (treatment x pho- toperiod, F(4, 165.01) = 5.00, p = 0.001) and a nested interaction (time of day within photope- riod, F(6, 209.40) = 22.04, p < 0.001) along with main effects of both treatment (F(4, 126.24) = 33.05, p < 0.001) and photoperiod (F(1, 192.72) = 522.51, p < 0.001). Moderate and high dose JZL increased NREM sleep time across the LP (t(145.25) 4.92, p < 0.001), while NREM sleep time was reduced on the recovery day during both the LP (t(145.26) = -3.36, p = 0.004) and DP (t(145.26) = -3.61, p = 0.002). Specifically, all three doses of JZL increase NREM sleep time during the first 3 Hr of the LP (ZT 00–03: t(274.85) 2.59, p 0.040) with the moderate dose increasing NREM sleep up to 6 Hr after administration (t(274.85) = 3.06, p = 0.010) and the high dose increasing NREM up to 9 Hr into the LP (t(274.85) = 2.52, p < 0.050). For NREM bout duration, there was an overall interaction (treatment x time of day within photoperiod; F(24, 218.31) = 1.67, p = 0.030), a secondary interaction (treatment x photope- riod; F(4, 120.79) = 2.80, p = 0.029), a nested interaction (time of day within photoperiod; F(6, 20.50) = 8.02, p < 0.001), and main effects of both treatment (F(4, 74.15) = 6.34, p < 0.001) and photoperiod (F(1, 179.88) = 132.73, p < 0.001). High and moderate dose JZL in- creased NREM bout duration across the LP (t(91.06) 2.94, p 0.016). The moderate dose of JZL increased NREM bout duration across the first 6 Hr of the LP (ZT00-06: t(225.33) 2.59, p 0.041), while high dose JZL only increased NREM bout duration later in the LP (ZT06-09: t(225.33) = 3.12, p = 0.008). The number of NREM bouts was not affected when JZL was administered before the LP. In contrast to the modest effects of JZL on REM sleep when the drug was given before the DP, administration of JZL before the LP produced a marked reduction in REM sleep (Fig 6C; bottom row). For the percent of time spent in REM sleep, there was an overall interaction (treatment x time of day within photoperiod, F(24, 224.83) = 1.61, p = 0.040), secondary inter- action (treatment x photoperiod, F(4, 116.88) = 13.58, p < 0.001), nested interaction (time of day within photoperiod, F(6, 212.50) = 9.60, p < 0.001), and main effects of both treatment (F

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 17 / 47 Endocannabinoid Signaling Regulates Sleep Stability

(4, 61.32) = 7.17, p < 0.001) and photoperiod (F(1, 200.12) = 10.77, p = 0.001). Both moderate and high dose JZL reduced REM sleep time across the LP (t(77.11) -4.81, p < 0.001). Specifi- cally, REM sleep was diminished across most time points in the LP following administration of both doses (moderate dose, ZT00-09: t(184.07) -2.81, p 0.022; high dose, ZT00-12: t (184.05) -3.87, p 0.001). For REM bout duration, there was an overall interaction (treat- ment x time of day within photoperiod, F(24, 192.48) = 2.02, p = 0.005), secondary interaction (treatment x photoperiod, F(4, 113.74) = 6.52, p < 0.001), and main effects of both treatment (F(4, 70.319) = 11.95, p < 0.001) and photoperiod (F(1, 165.44) = 15.90, p < 0.001). REM bout duration was suppressed by both moderate and high dose JZL across the LP (t(85.623) -27.46, p 0.001). Specifically, the moderate dose of JZL reduced REM bout duration during the first 6 Hr of the LP (ZT00-06: t(217.98) −3.27, p 0.005) and increased REM bout dura- tion during the middle of the DP (ZT18-21: t(217.49) = 2.75, p = 0.026). High dose JZL reduced REM bout duration across the entire LP and into the first 3 Hr of the DP (ZT00-15: t(214.55) −3.21, p 0.006). For the number of REM bouts, there was a secondary interaction (treatment x photoperiod, F(4, 119.91) = 10.01, p < 0.001), nested interaction (time of day within photope- riod, F(6, 206.20) = 11.17, p < 0.001), and main effects of both treatment (F(4, 74.09) = 3.14, p = 0.019) and photoperiod (F(1, 177.54) = 27.29, p < 0.001). Again, moderate and high dose JZL reduced the number of REM bouts across the LP (t(90.894) -3.41, p 0.002). The number of REM bouts was reduced at multiple time points during the LP following JZL administration (moderate dose, ZT03-06: t(227.24) = 3.19, p < 0.001); high dose, ZT00-09: t(227.24) -2.85, p 0.019). Thus, REM sleep is markedly suppressed by acute augmentation of 2-AG tone, but only when this drug is administered immediately before the time of day when mice engage in most of their REM sleep. EEG Power Spectral Measurements. Given the effects of increased 2-AG signaling on NREM and REM sleep described above, we examined the spectral content of the EEG signal from the experiment where JZL was administered before the DP (Fig 7). Similar results were obtained when JZL was administered before the LP (S4 Fig) and when CP47 was administered before the DP (S5 Fig). Despite the robust effects on sleep, JZL produced relatively modest effects on 12 Hr averages of EEG power spectrum from epochs of any state (Fig 7A–7C). To quantify JZL’s effects on EEG power spectra with higher temporal precision, we summed across well-described power spectral bandwidths (delta: 0-4Hz, theta: 4–8 Hz, and gamma: 30–70 Hz) in 3 Hr time bins (Fig 7D–7F). These bandwidths are routinely associated with sleep homeosta- sis (delta [43, 44], theta [45]), pneumonic processes (theta [46]), and attention (gamma [47]). Treatment with JZL had no effect on delta, theta, or gamma power during wake epochs (Fig 7D). For NREM epochs (Fig 7E), there was no effect of JZL on delta power, but for theta power there was a significant overall interaction (treatment x time of day within photoperiod, F(24, 335.61) = 1.84, p = 0.010), a nested interaction (time of day within photoperiod, F(6, 304.79) = 9.24, p < 0.001), and a main effect of photoperiod (F(1, 159.84) = 85.90, p < 0.001). However, there were no specific time points where JZL significantly altered NREM theta power relative to vehicle. For NREM gamma power, there was an overall interaction (treatment x time of day within photoperiod, F(24, 344.26) = 3.21, p < 0.001), a secondary interaction (treatment x pho- toperiod, F(4, 354.88) = 14.62, p < 0.001), a nested interaction (time of day within photoperiod, F(6, 304.49) = 25.78, p < 0.001), and main effects of both drug treatment (F(4, 220.36) = 10.85, p < 0.001) and photoperiod (F(1, 168.31) = 5.16, p = 0.024). JZL184 produced a dose-depen- dent reduction of NREM gamma power, with 8.0 mg/kg JZL184 decreasing gamma during the first 9 Hr of the DP (ZT 12–21: t(240.60) -2.67, p 0.032) and 16.0 mg/kg JZL reducing NREM gamma across the entire DP (ZT 12–00: t(159.50) -4.14, p 0.001). NREM gamma was no different from vehicle following the 1.6 mg/kg dose or on the recovery day. For REM epochs (Fig 7F), there was an there was an overall interaction (treatment x time of day within

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 18 / 47 Endocannabinoid Signaling Regulates Sleep Stability

Fig 7. MAGL Inhibition with JZL184 Attenuates Gamma Frequency Oscillations During Sleep. A-C, Average power spectra for epochs of different vigilance states across the entire DP (left hand) and LP (right hand). Solid lines denote means and shaded region around lines denotes SEM. A, Wake. B, NREM. C, REM. D-F, Change over the day in summated power in different frequency bandwidths from the power spectra: delta (left hand column), theta (middle column), and gamma (right hand column). In these graphs, results from epochs of wake are denoted in red (D), NREM are in blue (E), and REM are in green (F). Symbols/Bars represent mean±SEM for 3 hr time bins (N = 10). Grey background in graphs shows dark photoperiod. Asterisks denote significant difference from vehicle baseline. All injections administered at onset of DP (ZT 12:00). doi:10.1371/journal.pone.0152473.g007

photoperiod, F(24, 284.87) = 1.71, p 0.022) and a main effect of photoperiod (F(1, 306.75) = 16.23, p < 0.001) for delta power. However, there were no pair-wise differences between treat- ment/recovery conditions and vehicle. Similarly, for REM theta power, there was an overall interaction (treatment x time of day within photoperiod, F(24, 293.65) = 2.36, p < 0.001) and a nested interaction (time of day within photoperiod, F(6, 292.49) = 8.09, p < 0.001), but there were no pair-wise differences between treatment/recovery conditions and vehicle. For REM gamma, there was a secondary interaction (treatment x photoperiod, F(4, 252.78) = 5.03, p = 0.001) and a nested interaction (time of day within photoperiod, F(6, 292.39) = 10.94, p < 0.001) with main effects of drug treatment (F(4, 83.77) = 7.39, p < 0.001) and photoperiod (F(1, 235.60) = 15.65, p < 0.001). JZL reduced REM gamma in a dose-dependent manner with 8.0 mg/kg JZL184 decreasing REM gamma across the first half of the DP (ZT12-18: t(121.00) -3.04, p 0.011) and 16.0 mg/kg JZL184 decreasing REM gamma across the entire DP (ZT12-00: t(85.92) -3.64, p 0.002). There was no difference in REM gamma power follow- ing low dose JZL184 or on the recovery day. Thus, increasing endogenous 2-AG tone with JZL has little effect on delta or theta power, but it attenuates gamma oscillations, particularly during

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 19 / 47 Endocannabinoid Signaling Regulates Sleep Stability

sleep. Additionally, this effect is consistent irrespective of the time of day JZL is administered (see S4 Fig).

Inhibition of Fatty Acid Amide Hydrolase Stabilizes NREM and Suppresses REM Sleep Measurements. Next, we tested the hypothesis that endogenous N-acylethanola- mines, including AEA, could modulate sleep. The FAAH inhibitor URB was injected systemi- cally at three different doses (0.1, 1.0, and 10.0 mg/kg) over successive days (S6 Fig, panel A), but URB did not have a substantial effect on either NREM (S6 Fig, panel B) or REM (S6 Fig, panel C) sleep, in contrast to the counterintuitive effects of i.c.v. injection reported previously [18]. These data would appear to suggest that N-acylethanolamines are not important for the regulation of vigilance states. However, application of exogenous AEA is known to facilitate NREM sleep [14, 15], and the elevation of N-acylethanolamines in rodent brain tissue by URB lasts only a few hours [48]. Thus, we performed a separate experiment with a single dose of the selective, long-lasting FAAH inhibitor AM3506 (10.0 mg/kg; Fig 8A) that reduces FAAH activ- ity for up to 10 days after administration [49]. In this experiment, subjects were administered a vehicle injection followed 24 Hr later by an injection of AM3506, and polysomnographic mea- sures of sleep were obtained over the following 48 Hr. As shown in Fig 8B, AM3506 signifi- cantly altered NREM sleep. For NREM sleep time, there was a significant overall interaction (treatment x time of day within photoperiod, F(18, 142.90) = 3.68, p < 0.001), a secondary interaction (treatment x photoperiod, F(2, 80.60) = 20.57), and main effects of both treatment (F(2, 56.53) = 11.63, p < 0.001) and photoperiod (F(1, 111.44) = 231.09, p < 0.001). During the DP, AM3506 significantly augmented NREM sleep time (t(66.28) = 5.16, p < 0.001), and simi- lar to the effect of JZL, there was a significant reduction in NREM sleep during the DP on the recovery day (t(66.63) = -2.41, p = 0.038). In contrast to the effects of JZL and CP47, NREM sleep time during the LP was unaffected. Pair-wise comparisons at individual time bins found AM3506 significantly increased NREM sleep across the first 9 Hr of the DP (ZT12-21: t (168.29) 2.64, p 0.018), and on the recovery day, NREM sleep time was significantly reduced during the first three hours of the DP (ZT12-15: t(168.35) = -3.25, p = 0.003). Thus, increasing N-acylethanolamine signaling with long-lasting inhibition of FAAH increases NREM sleep time, but this does not produce the biphasic effect seen with JZL and CP47. To ascertain the effect of FAAH inhibition on NREM architecture, we measured the number and duration of NREM bouts at 3 Hr time points across the circadian cycle (Fig 8B). There was an overall interaction for NREM bout duration (treatment x time of day within photoperiod, F (18, 148.23) = 2.79, p < 0.001) and a secondary interaction between treatment and photoperiod (F(2, 75.68) = 3.54, p = 0.034). There was a significant increase in NREM bout duration during the second quarter of the dark photoperiod (ZT15-18: t(135.38) = 2.77, p = .013). The number of NREM bouts was not affected by treatment with AM3506. These findings demonstrate that FAAH inhibition promotes sleep by increasing NREM stability shortly after drug administration. Similar to JZL, AM3506 reduced REM sleep (Fig 8C). For the percent time spent in REM, there was a nested interaction (time of day within photoperiod, F(6, 139.71) = 7.51, p < 0.001) and main effects of treatment (F(2, 51.93) = 4.399, p = 0.017) and photoperiod (F(1, 112.07) = 227.69, p < 0.001). Overall, AM3506 reduced REM sleep (t(47.65) = -2.75, p = 0.017), specifi- cally during the third quarter of the DP (ZT18-21: t(158.67) = -2.54, p = 0.024) and first quarter of the LP (ZT00-03: t(158.67) = -3.05, p = 0.005). For the duration of REM bouts, there was a nested interaction (time of day within photoperiod, F(6, 132.64) = 3.99, p = 0.001) and main effects of treatment (F(2, 59.13) = 10.66, p < 0.001) and photoperiod (F(1,105.96) = 7.72,

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 20 / 47 Endocannabinoid Signaling Regulates Sleep Stability

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 21 / 47 Endocannabinoid Signaling Regulates Sleep Stability

Fig 8. FAAH Inhibition with AM3506 Increases NREM Sleep Time and Stability while Decreasing REM Sleep. A, Diagram of experimental protocol for recording sleep after administration of the long-lasting FAAH inhibitor AM3506. B, Quantification of NREM sleep time and architecture for the AM3506 experiment (N = 9). C, Quantification of REM sleep time and architecture. In all graphs, the grey shaded region denotes the dark photoperiod. Symbols represent mean±SEM for 3 Hr time bins. Asterisks denote significant difference from vehicle baseline. All injections administered at onset of dark photoperiod (ZT 12:00). doi:10.1371/journal.pone.0152473.g008

p = 0.006). Overall, REM bouts were longer on the recovery day (t(60.507) = 2.74, p = 0.016), but AM3506 reduced REM bout duration across the DP (t(73.75) = -2.70, p = 0.017), specifi- cally during the middle of the DP (ZT15-21: t(170.06) -2.61, p 0.020). Finally, for the number of REM bouts, there was a nested interaction (time of day within photoperiod, F(6, 142.95) = 5.23, p < 0.001) and main effects of treatment (F(2, 46.33) = 4.39, p = 0.018) and photoperiod (F(1, 120.03) = 84.76, p < 0.001). The number of REM bouts during the LP was reduced by AM3506 and on the recovery day (t(55.64) -2.80, p 0.014), specifically during the third quarter of the LP (ZT06-09: t(139.96) -2.68, p 0.017). EEG Power Spectral Measurements. The results of power spectral analysis of the EEG from the AM3506 experiment were very similar to those obtained following CP47 and JZL administration (S7 Fig). Specifically, increasing AEA tone with AM3506 had modest effects on delta and theta bandwidths, but it reduced gamma power during NREM and REM epochs. Thus, the attenuation of gamma oscillations, particularly during sleep, seems to be a consistent effect of increased eCB signaling.

Blockade of CB1 Fragments NREM Sleep and Substantially Alters Power Spectral Features of the EEG Sleep Measurements. To determine if eCB/CB1 signaling is necessary for the normal cir- cadian fluctuation in NREM and REM sleep, we performed experiments with the full, selective CB1 antagonist/inverse agonist AM281. Following AM281 administration, there was a sub- stantial fragmentation of NREM sleep and a loss of REM, particularly when this drug was administered prior to the LP (Fig 9). Again, given that eCBs exhibit a circadian fluctuation that differs by brain region, it was not easy to predict a priori an optimal time to administer the drug, so two separate experiments were performed where AM281 was given at opposite points in the circadian phase, immediately before either the LP or DP. In both experiments, two doses of AM281, 0.5 mg/kg (low) and 5.0 mg/kg (high), were administered sequentially on consecu- tive days following a baseline day when vehicle was given (Fig 10A). When AM281 was administered prior to the DP (Fig 10B, top row), NREM sleep time was affected by an overall interaction (treatment x time of day within photoperiod, F(18, 198.82) = 10.27, p < 0.001) with a main effect of photoperiod (F(1, 169.19) = 836.77, p < 0.001). Only the high dose of AM281 produced effects on NREM sleep time, and the magnitude of these effects was small. Specifically, there was an increase in NREM sleep time during the first quarter of the DP (t(261.07) = 2.93, p = 0.007) and a decrease during the second quarter (t(261.09) = -3.20, p = 0.003). The effect on NREM bout duration was more pronounced with a significant overall interaction (treatment x time of day within photoperiod, F(18, 201.36) = 7.20, p < 0.001) and main effects of both treatment (F(2, 69.38) = 3.86, p = 0.026) and photope- riod (F(1, 163.20) = 86.17, p < 0.001). Following administration of high dose AM281, there was an overall reduction in NREM bout duration (t(72.79) = -2.64, p = 0.020) with a significant reduction during the second quarter of the DP (ZT15-18: t(205.68) = -3.02, p = 0.006). The number of NREM bouts was also affected by an overall interaction (F(18, 204.63) = 3.46, p < 0.001) with main effects of both treatment (F(2, 61.58) = 3.30, p = 0.043) and photoperiod (F(1, 173.11) = 25.90, p < 0.001). However, this result was largely driven by an increased

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 22 / 47 Endocannabinoid Signaling Regulates Sleep Stability

Fig 9. Example EEG/EMG Traces on Different Time Scales Following Vehicle or AM281 Administration. EEG and EMG traces are from the same subject at the same stage of the circadian cycle after administration of either vehicle (A, A’, and left column of C) or 5 mg/kg AM281 (B, B’, and right column of C). Data are from experiment with AM281 administration before the LP. Panels A and B show a 2 Hr 15 min window from ZT 00:15–02:30, roughly 15–30 min after drug administration, coinciding with peak effects observed on sleep. Panels A’ and B’ show a 15 min long segment expanded from the region in A and B highlighted by the dashed orange box. Panel C shows representative 18 sec long data segments corresponding to NREM and wake obtained following vehicle and AM281 administration. These data segments were taken from the segments shown in A and B. The color-coded hypnogram shown at the bottom of A, B, A’, and B’ represents consecutive 2 sec epochs shown as wake (red), NREM (blue), REM (green), and unclassified (grey). Note the loss of REM sleep and fragmentation of NREM following AM281 administration. Black traces depict EEG, red traces depict EMG. A and B are identically scaled. A’ and B’ are identically scaled. All traces in C are identically scaled. doi:10.1371/journal.pone.0152473.g009

number of NREM bouts during the first quarter of the DP following high dose AM281 (ZT612-15: t(181.59) = 4.81, p < 0.001). Thus, administration of AM281 prior to the DP yields subtle effects on sleep time but decreases the stability of NREM bouts. REM sleep was disrupted by AM281, but the effect size was small as baseline REM is normally very low during the DP, when mice are most active (Fig 10B, bottom row). For the percent of time spent in REM, there was a nested interaction (time of day within photoperiod, F(6, 187.08) = 16.15, p < 0.001) and main effects of both treatment (F(2, 85.82) = 4.28, p =0.017)andphotope- riod (F(1, 149.93) = 377.93, p < 0.001). Overall, 5 mg/kg AM281 decreased REM sleep (t(89.77) = -2.92, p = 0.009), particularly during the third quarter of the DP (ZT18-21: t(241.64) = -2.88, p = 0.009). REM bout duration was affected by an overall interaction (treatment x time of day

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 23 / 47 Endocannabinoid Signaling Regulates Sleep Stability

Fig 10. Blockade of CB1 Has Minimal Effects on NREM Sleep Time but Fragments NREM, Resulting in Reduced REM Sleep. A, Schematic overview of experimental paradigm. B, Effect of administering different doses of AM281 at the onset of the DP (ZT 12:00, N = 12). C, Effect of administering different doses of AM281 at the onset of the LP (ZT 00:00, N = 9). B&C,Measures of NREM (top row) and REM (bottom row) sleep time and architecture are shown. In all graphs, grey shaded regions denote the the DP. Asterisks (*) denote significant pair-wise comparisons between drug conditions and measures obtained during vehicle baseline. Symbols represent means±SEM across all subjects for each 3 Hr time bin. doi:10.1371/journal.pone.0152473.g010

within photoperiod, F(12, 176.73) = 2.60, p = 0.003), nested interaction (time of day within photo- period, F(6, 169.25) = 6.37, p < 0.001), and main effect of photoperiod (F(1, 151.57) = 24.16, p < 0.001). The 0.5 mg/kg dose of AM281 increased REM bout duration during the second quar- ter of the DP (ZT15-18: t(179.07) = 3.04, p = 0.005). There was no effect of AM281 on the number of REM bouts. When AM281 was administered prior to the LP, overall sleep time did not change substan- tially, but there was a profound fragmentation of NREM (Fig 10C, top row). For NREM sleep time, there was an overall interaction (treatment x time of day within photoperiod, F(18, 146.08) = 8.76, p < 0.001), a secondary interaction (treatment x photoperiod, F(2, 102.87) = 3.31, p = 0.040) and main effects of both treatment (F(2, 81.59) = 3.97, p = 0.023) and photope- riod (F(1, 122.15) = 383.92, p < 0.001). When delivered at the onset of the LP, AM281 increased overall NREM sleep time (t(84.01) = 2.74, p = 0.015), but a comparison between pho- toperiods found that NREM sleep time was only increased during the DP (low dose: t(91.65) = 2.38, p = 0.039; high dose: t(91.65) = 2.84, p = 0.011). More specifically, there was a significant

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 24 / 47 Endocannabinoid Signaling Regulates Sleep Stability

increase in NREM following high dose AM281 during the first 3 Hr of the DP (ZT12-15: t (191.99) = 3.15, p = 0.004), which was the same point in the circadian cycle when NREM sleep time was increased in the experiment where AM281 was delivered before the DP. For NREM bout duration, there was an overall interaction (treatment x time of day within photoperiod, F (18, 141.98) = 4.74, p < 0.001), a secondary interaction (treatment x photoperiod, F(2, 82.68) = 12.61, p < 0.001), and main effects of both treatment (F(2, 59.88) = 9.86, p < 0.001) and photo- period (F(1, 109.96) = 31.83, p < 0.001). High dose AM281 substantially reduced NREM bout duration (t(62.07) = -4.42, p < 0.001), particularly during the LP (t(69.68) = -6.23, p < 0.001). More specifically, NREM bout duration was reduced for the first 3 Hr of the LP following low dose AM281 (ZT00-03: t(176.12) = -2.82, p = 0.011) and for the first 9 Hr following high dose AM281 (ZT00-09: t(176.12) -3.46, p 0.001). The number of NREM bouts was affected in the opposite manner. There was an overall interaction (treatment x time of day within photo- period, F(18, 144.093) = 3.266, p < 0.001), a secondary interaction (treatment x photoperiod, F (2, 78.77) = 13.65, p < 0.001), and main effects of both treatment (F(2, 53.14) = 19.99, p < 0.001) and photoperiod (F(1, 113.94) = 148.145, p < 0.001). High dose AM281 increased the number of NREM bouts (t(63.61) = 6.79, p < 0.001) particularly during the first 9 Hr of the LP (ZT00-09: t(159.32) 4.22, p < 0.001), and both doses increased the number of NREM bouts during the second quarter of the DP (ZT15-18: t(159.32) 2.31, p 0.045). Thus, block- ade of CB1 receptors greatly fragments NREM sleep, but opposing effects on NREM bout dura- tion and the number of NREM bouts result in subtle changes in total sleep time. In addition to the effects on NREM sleep, REM sleep time was significantly reduced follow- ing AM281 administration prior to the LP (Fig 10C, bottom row). For the percent of time spent in REM sleep, there was a secondary interaction (treatment x photoperiod, F(2, 95.51) = 36.30, p < 0.001), nested interaction (time of day within photoperiod, F(6, 143.275) = 11.15, p < 0.001), and main effects of both treatment (F(2, 74.2) = 21.38, p < 0.001) and photoperiod (F(1, 116.42) = 68.40, p < 0.001). Both low and high dose AM281 reduced REM sleep during the LP (t(84.01) -4.22, p < 0.001). Low dose AM281 reduced REM sleep during the first 6 Hr of the LP (ZT00-06: t(190.28) -3.30, p 0.002), and high dose AM281 reduced REM at all times during the LP (ZT00-12: t(190.28) -3.40, p 0.002). For REM bout duration, there was a secondary interaction (treatment x photoperiod, F(2, 81.71) = 7.10, p = 0.001), nested interaction (time of day within photoperiod, F(6, 139.18) = 5.34, p < 0.001), and a main effect of treatment (F(2, 56.07) = 3.35, p = 0.042). The high dose of AM281 reduced REM bout duration during the LP (t(64.52) = -4.05, p < 0.001), particularly during the first 9 Hr of the LP (ZT00-09: t(156.05) -2.31, p 0.044). For the number of REM bouts, there was an overall interaction (treatment x time of day within photoperiod, F(12, 133.54) = 1.85, p = 0.046), secondary interaction (treatment x photoperiod, F(2, 68.89) 7.463, p = 0.001), nested interaction (time of day within photoperiod, F(6, 127.15) = 3.99, p = 0.001), and main effects of both treatment (F(2, 46.46) = 6.39, p =0.004) and photoperiod (F(1, 99.89) = 59.86, p < 0.001). The number of REM bouts were reduced by high dose AM281 during the LP (t(54.99) = -4.49, p < 0.001) and increased by low dose AM281 during the DP (t(54.65) = 2.44, p = 0.036). At the high dose, AM281 reduced the number of REM bouts across most of the LP (ZT00-06 & 09–12: t(162.69) -2.67, p = 0.002), while low dose AM281 decreased the number of REM bouts during the second quarter of the LP (ZT03-06: t (162.62) = -3.30, p = 0.002) and increased REM bouts during the third quarter of the DP (ZT18- 21: t(157.68) = 2.51, p = 0.026). Thus, blockade of CB1 signaling fragments NREM and decreases REM sleep, suggesting that this receptor is necessary for NREM stability. EEG Power Spectral Measurements. Given that blockade of endocannabinoid signaling through CB1 fragments NREM sleep, we hypothesized that power spectral features associated with sleep might be disrupted after acute administration of CB1 antagonists. Administration of AM281 before the LP had large effects on power spectral features of the EEG across vigilance

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 25 / 47 Endocannabinoid Signaling Regulates Sleep Stability

states, but the nature of these effects was different across states (Fig 11A–11C). Notably, the power of low frequencies (< 8 Hz) was consistently increased, and high frequencies (gamma, > 30 Hz) were much less affected by CB1 blockade. These effects lasted for most of the day, and a similar time course was observed in experiments where AM281 was administered before the DP (S8 Fig), suggesting that this effect is not modulated by circadian processes. As for we did for CP47, AM3506, and JZL184 we quantified power spectral bandwidths (delta, theta, and gamma) in 3 Hr time bins over the entire recording (Fig 11D–11F). During wake epochs (Fig 11D) there was a significant interaction (treatment x time of day within pho- toperiod, F(24, 229.63) = 5.26, p < 0.001) with main effects of both treatment (F(3, 73.99) = 42.19, p < 0.001) and photoperiod (F(1, 154.27) = 127.39, p < 0.001) for delta power. Only the 5.0 mg/kg dose significantly elevated wake delta relative to vehicle (t(53.52) = 7.21, p < 0.001), and comparisons at individual time points found that this effect lasted for 18 Hr after drug administration (ZT00-18: t(83.70) > 3.13, p 0.007). Wake theta power was also modulated by a significant interaction (F(24, 228.72) = 3.23, p < 0.001) and main effects of both treatment (F(3, 69.952) = 20.74, p < 0.001) and photoperiod (F(1, 157.78) = 13.20, p < 0.001). Again, only the high dose of AM281 significantly elevated theta power over the circadian cycle (t (50.532) = 5.35, p < 0.001), and theta power was increased over the first 18 Hr of the recording period (ZT00-18: t(80.45) > 3.97, p < 0.001). Analysis of wake gamma power also found an overall interaction (F(24, 227.65) = 8.013, p < 0.001) with a main effect of photoperiod (F(1, 143.48) = 89.70, p < 0.001), but no pair-wise comparisons between treatment conditions and the vehicle baseline reached significance. AM281 also altered EEG power spectra during NREM epochs (Fig 11E). For NREM delta power, there was a significant overall interaction (F(24, 229.40) = 9.84, p < 0.001) with a main effect of treatment (F(3, 80.45) = 28.89, p < 0.001). 5 mg/kg AM281 produced an overall increase in NREM delta power (t(58.23) = 5.54, p < 0.001) with pair-wise differences noted across the vast majority of the recording (ZT00-21: t(88.09) 2.53, p 0.039). There was also an overall interaction for NREM theta power (F(24, 235.54) = 6.31, p < 0.001) with main effects of both treatment (F(3, 148.62) = 35.06, p < 0.001) and photoperiod (F(1, 135.42) = 5.438, p = 0.21). 5 mg/kg AM281 increased the overall power in the theta bandwidth during NREM sleep (t(118.00) = 5.01, p < 0.001) with specific pair-wise comparisons over the major- ity of the recording (ZT00-21: t(150.13) 2.86, p 0.014). Finally, analysis of NREM gamma found an overall interaction (F(24, 233.59) = 14.55, p < 0.001) with main effects of both treat- ment (F(3, 121.33) = 7.128, p < 0.001) and photoperiod (F(1, 127.916) = 93.21, p < 0.001), but there were no significant differences between AM281 and vehicle. The CB1 antagonist also affected EEG spectral content during REM (Fig 11F). For REM delta power, there was an overall interaction (F(24, 217.83) = 1.68, p = 0.028) with a main effect of treatment (F(3, 54.98) = 7.64, p < 0.001). 5.0 mg/kg AM281 increased delta power during REM sleep epochs across the day (t(40.91) = 2.82, p < 0.022), but this was mainly due to increased REM delta during the first 9 Hr (ZT00-09: t(76.78) 3.54, p 0.002). For REM theta power, there was a secondary interaction (treatment x photoperiod, F(3, 146.49) = 9.23, p < 0.001) with main effects of both treatment (F(3, 51.66) = 19.22, p < 0.001) and photope- riod (F(1, 219.58) = 18.05, p < 0.001). Overall, 5.0 mg/kg AM281 increased REM theta power (t(39.91) = 5.23, p < 0.001), particularly during the first 15 Hr of the recording (ZT00-15: t (85.58) 3.92, p 0.001). There was a treatment x photoperiod interaction for REM gamma power (F(3, 168.32) = 3.61, p = 0.015) with a main effect of photoperiod (F(1, 196.66) = 8.13, p = 0.005). Overall, REM gamma was augmented by the high dose of AM281 during the light photoperiod (t(51.71) = 3.12, p = 0.009), but there were no differences at specific time points. Thus, blockade of CB1 receptors produces broadband changes in the EEG waveform that are particularly evident in lower frequencies irrespective of vigilance state.

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 26 / 47 Endocannabinoid Signaling Regulates Sleep Stability

Fig 11. Blockade of CB1 Receptors Produces Broadband Changes in EEG Power Spectral Features. Results are from experiment where AM281 was administered at the onset of the LP. A—C, Average of power spectra for each photoperiod and for each vigilance state across different days of the experiment. Solid lines indicate mean of all subjects as a function of frequency, and shaded regions surrounding lines denotes standard error of the mean. A, Power spectra from wake epochs. B, Power spectra from NREM epochs. C, Power spectra from REM epochs. D–F, Average power in specified bandwidths in each state for 3 Hr epochs over the day. Data from each vigilance state are color coded with wake in red (D), NREM in blue (E), and REM in green (F). Gray backgrounds indicate the DP. Asterisks (*) denote significant pair-wise comparisons between drug conditions and measures obtained during vehicle baseline. Symbols represent means±SEM across all subjects (N = 9) for each 3 Hr time bin. doi:10.1371/journal.pone.0152473.g011 eCB Signaling is Not Necessary for Sleep Homeostasis but is Required for the Stability of Rebound Sleep Measures of EEG delta and theta power are frequently used as an index of sleep drive [44, 50, 51], and according to this interpretation, the large increase in NREM delta power following administration of AM281 may be indicative of augmented sleep homeostatic drive. Thus, we sought to test this possibility by administering AM281 following 6 Hr of total sleep deprivation (TSD) using a rotating disc paradigm (Fig 12A & S1 Fig). Measurements of sleep during the dep- rivation period clearly show that this method was successful in achieving TSD for 6 Hr (Fig 12B). Examination of the percent of time subjects spent in NREM sleep (Fig 12C, top) found an overall interaction (group x time of day within photoperiod within experimental phase, F(24, 213.06) = 10.34, p < 0.001) with main effects of experimental phase (F(1, 112,24) = 8.26, p < 0.005) and photoperiod (F(1, 165.42) = 187.28, p < 0.001). In general, subjects engaged in more NREM sleep following sleep deprivation (t(112.24) = 2.88, p = 0.005), indicating a signifi- cant NREM rebound. Compared to baseline, the vehicle group spent significantly more time in

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 27 / 47 Endocannabinoid Signaling Regulates Sleep Stability

Fig 12. Endocannabinoid Signaling Is Necessary for the Rebound in NREM Duration Following Sleep Deprivation, but Blockade of CB1 Has only a Weak Effect on the Rebound in Total Sleep Time. A, Schematic overview of sleep deprivation experimental paradigm. B, Time course of NREM sleep time across all days of sleep deprivation experiment. Horizontal red bar on the Sleep Dep day indicates when sleep deprivation took place. Vertical dotted lines denote boundaries between experimental phases. Blue symbols/lines show AM281 group (N = 9) while the vehicle group (N = 11) is depicted in black. Downward facing arrows () indicate time of drug administration. C, Comparisons of total NREM sleep time (top), NREM bout duration (middle), and the number of NREM bouts (bottom) on the first baseline day (vehicle administration all groups) and first day of recovery (vehicle or AM281 administration). Asterisks (*) denote significant pair-wise comparisons within-groups between drug conditions and measures obtained during vehicle baseline. Daggers (†) denote significant pair-wise comparisons between groups on the recovery day. In B & C, Grey shaded regions indicate the DP, and symbols/bars represent means±SEM across all subjects for each 3 Hr time bin. doi:10.1371/journal.pone.0152473.g012

NREM during the first 3 Hr of recovery (ZT06-09: t(286.99) = 2.40, p = 0.017) and during the first 3 Hr of the DP (ZT12-15: t(286.99) = 3.08, p = 0.002). Similarly, the AM281 group had increased NREM sleep during the first 3 Hr of the DP (ZT12-15: t(286.99) = 1.99, p = 0.47). There were no differences in the amount of NREM sleep between the vehicle and AM281 treat- ment groups during either baseline or recovery. Additionally, there was no difference between treatment groups in the recovery of the sleep deficit induced by the 6 Hr TSD protocol (S9 Fig). Thus, both groups displayed an augmentation of NREM sleep time following 6 Hr sleep depri- vation, and the two groups did not differ in regards to the total amount of sleep obtained dur- ing recovery from sleep deprivation.

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 28 / 47 Endocannabinoid Signaling Regulates Sleep Stability

In contrast, there were significant between-group differences for NREM architecture during recovery from sleep deprivation. For NREM bout duration (stability; Fig 12C, middle), there was an overall interaction (group x time of day within photoperiod within experimental phase, F(24, 213.74) = 8.66, p < 0.001) with main effects of treatment group (F(1, 76.81) = 7.14, p = 0.009) and photoperiod (F(1, 158.63) = 37.26, p < 0.001). The AM281 group had reduced NREM stability compared to the vehicle group (t(76.81) = -2.67, p = 0.009), with specific differ- ences evident at two time points of recovery (ZT06-09: t(256.30) = -3.63, p < 0.001; ZT00-03: t(256.30) = -2.47, p = 0.014). There were no between-group differences during the baseline recordings. Relative to their own baseline, the vehicle group had increased NREM stability during the first 3 Hr of recovery (ZT06-09: t(265.32) = 3.95, p < 0.001), but the AM281 treated mice did not show increased NREM stability following TSD. Rather, the AM281 group had decreased NREM stability during the second half of the recording (ZT18-21 & 00–03: t(265.32) -1.99, p 0.048). For the number of NREM bouts (Fig 12C, bottom), there was an overall interaction (group x time of day within photoperiod within experimental phase, F(24, 220.93) = 2.95, p < 0.001) and a main effect of photoperiod (F(1, 174.17) = 17.77, p < 0.001). There were no differences between groups during baseline, but the AM281 group had significantly more NREM bouts than the vehicle group during the first three hours of recovery (ZT06-09: t(169.85) = 2.16, p = 0.032). Relative to their own baseline, the vehicle group increased the number of NREM bouts during the first quarter of the DP (ZT12-15: t(177.13) = 2.20, p = 0.029). These findings show that AM281 blocks the increase in NREM sta- bility during recovery from TSD and increases the number of NREM bouts to compensate. REM sleep was also affected by TSD with reduced REM early in recovery in both the AM281 and vehicle groups (S10 Fig). Late in the recovery, there was increase in REM sleep in the AM281 treated group due to an increase in the number of REM bouts. Consistent with our previous experiments, blockade of CB1 receptors produced a large increase in low frequency oscillations that occluded any changes due to sleep deprivation (S11 Fig). However, wake theta power did increase during the sleep deprivation period (S11 Fig, panel B). Thus, power spectral results cannot be used as an index of homeostatic drive in this experiment as they are con- founded by the baseline effects of CB1 antagonism.

Discussion The major findings of this work are that eCB signaling through the CB1 receptor is necessary and sufficient for the stability of NREM sleep. Direct activation of CB1 with CP47 or increasing eCB tone with JZL or AM3506 augmented the time spent in NREM primarily due to increased NREM bout length. This suggests that increasing eCB signaling stabilizes the NREM state. Notably these effects were biphasic, where the initial increase in sleep and NREM stability gave way to a secondary reduction and destabilization of NREM bouts suggesting that homeostatic regulation of sleep remained intact. Further support for the role of eCBs in regulating NREM stability comes from experiments with the CB1 antagonist AM281, where blockade of CB1 reduced the duration of NREM bouts. Importantly, this effect was largely offset by a concomi- tant increase in the number of NREM bouts resulting in minimal changes in overall sleep time. Again, the opposing effects of reduced NREM bout duration and number suggest homeostatic regulation of sleep remains intact in the absence of CB1 signaling. However, CB1 blockade con- sistently resulted in substantial broadband changes to EEG power spectra, particularly in low frequency bandwidths that are often used as correlates of sleep drive. To test the involvement of eCB signaling in sleep homeostatic processes, we administered AM281 immediately prior to a recovery from TSD. Blockade of CB1 during recovery from TSD did not occlude the rebound in NREM sleep time following TSD, but it did completely block the increased stability (bout

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 29 / 47 Endocannabinoid Signaling Regulates Sleep Stability

duration) of NREM sleep. Unfortunately, the baseline effects of CB1 antagonism on EEG power spectra hobbled our attempts to use this as a metric of sleep drive. Thus, the major con- clusion from this work is that eCBs regulate sleep stability, but eCB signaling is not necessary for sleep homeostasis.

Validity and Robustness of State-Scoring Algorithm In order to perform the series of experiments described herein, we needed to find a way to score mouse polysomnographic data in an accurate and timely manner, so we developed and validated the state-space based approach. Our method was heavily influenced by previous reports, but we modified and extended this work, in part by automating the state-assignment process after state-space coordinates have been defined. Discrete fourier analysis is the basis of this approach, because the state-space coordinates are derived from power spectral ratios simi- lar to those used by others [36, 38, 39, 52, 53]. The use of power spectral ratios (effectively nor- malizing the FFT) to define state-space coordinates, as opposed to raw power spectral values [54], means that our method of state-scoring is robust to changes in the FFT, and this is evi- denced by our experimental results. Activating CB1/eCB signaling with CP47, JZL, or AM3506 produced only modest effects on power spectral measurements, but these drugs had large biphasic effects on measures of sleep time. Importantly, the effects of these drugs on sleep and EEG power spectra were not temporally aligned. In contrast, CB1 blockade consistently pro- duced broadband changes in EEG power spectra that were most evident for low frequency bandwidths, and these effects were evident irrespective of the time of day of drug administra- tion or if CB1 was blocked following TSD. However, CB1 antagonism did not produce substan- tial effects on total sleep time and the magnitude of effects on sleep architecture varied with time of drug administration. Thus, while CB1 antagonism consistently and substantially alters EEG power spectra, the time course of these effects are not aligned with changes in the sleep score or even sleep architecture. Consequently, the various data sets presented here show large changes in sleep in the absence of substantial changes in state-dependent EEG power spectra (direct and indirect CB1 agonists) and large changes in power spectra in the absence of sub- stantive changes in sleep time (CB1 antagonism). Therefore, we conclude that in addition to performing comparably to human scoring, our vigilance state scoring algorithm is robust to cannabinoid pharmacological manipulations that alter EEG power spectra.

The Biphasic Effects of Augmented eCB Signaling The early facilitation of NREM sleep seen in the present study following activation of CB1 is largely in agreement with earlier findings [6–11, 13–15, 29, 32], but the secondary (wake pro- moting/sleep fragmenting) effect of CP47 and JZL to reduce NREM sleep at a time of day when mice normally engage in most of their daily sleep is a novel finding. It should be mentioned that JZL had lower magnitude effects (change from vehicle) on NREM when it was adminis- tered before the LP, and did not produce substantial fragmentation of sleep during the subse- quent DP. It is unclear why a secondary process was not evident in this experiment, but the acute augmentation of sleep stability was, in this case, synchronized with the normal time of day when mice are inactive. Thus, there was no perturbation in the circadian timing or homeo- static regulation of NREM and wake as in experiments where JZL, CP47, and AM3506 were administered before the DP. Based on several observations, we hypothesize that the sleep fragmentation during the sec- ondary phase of CP47, JZL, and AM3506 effects is due to a rapid reduction in CB1 signaling. First, JZL selectively increases 2-AG levels via inhibition of MAGL, and both CP47 and 2-AG are highly potent, full agonists at the CB1 receptor [55–57]. In contrast, as an irreversible

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 30 / 47 Endocannabinoid Signaling Regulates Sleep Stability

inhibitor of FAAH, AM3506 increases N-acylethanolamine transmitters for several days [49], and AEA is only a weakly efficacious agonist at CB1 [58]. This discrepancy could explain the difference in the magnitude of effects seen between these drugs. On the other hand, the lack of any effect of URB597 even during the first few hours of the recording is surprising, and it is possible that the early induction of sleep by AM3506 could be due to off-target augmentation of 2-AG. However, administration of exogenous AEA increases NREM sleep [13, 14], and administration of THC, a partial agonist at CB1, also increases NREM sleep (unpublished observation). Thus, partial activation of CB1 with endogenous N-acylethanolamines is likely to have hypnogenic effects. Second, following acute activation, CB1 receptors are rapidly down- regulated on a time scale compatible with the biphasic effects of CP47 and JZL [59, 60]. As these compounds are slowly eliminated in the hours following injection, receptor function will be decreased compared to baseline. Although data on diurnal expression of CB1 are extremely limited, a few studies have provided evidence that CB1 accumulates in brain and neuroendo- crine tissue during the DP [61, 62]. Thus, it is possible that diurnal fluctuations in CB1 expres- sion could account for the blunted biphasic effects seen with JZL administration before the LP. Third, NREM fragmentation was also observed in the present study following administration of CB1 antagonists, and CB1 knockout mice also have fragmented sleep [24, 25].

On the Timing of Drug Administration Although CB1 blockade consistently reduced NREM bout duration, the effects were greatest when AM281 was administered at the onset of the LP. Similarly, JZL had differential effects on sleep depending on the circadian timing of drug administration. When JZL was administered either before the LP or the DP, it increased NREM sleep time and bout duration, but the mag- nitude of these effects was blunted when the drug was given before the LP. Consequently, the timing of maximal effects for activation and inhibition of eCB signaling peaks at opposite poles of the circadian cycle, suggesting that eCB signaling interacts with or is controlled by circadian processes. These are not mutually exclusive possibilities, and there is evidence supporting both mechanisms. First, cannabinoids alter the light-induced phase-shift in activity in free running rodents [63, 64], suggesting that CB1 can influence entrainment to photic stimulation. Addi- tionally, eCB levels fluctuate over the circadian period in a brain-region dependent manner [33, 34], suggesting that eCB signaling may be under circadian regulation. In fact, levels of AEA and 2-AG are lowest in cortical and hypothalamic tissues at the onset of the DP [33], and in the present work, administration of CB1 antagonists at this time point produced smaller effects on sleep architecture. Thus, our findings provide evidence that changing levels of the molecular components of the eCB system have functional consequences for eCB regulation of vigilance states.

eCBs and EEG Power Spectra NREM delta power has been used as an index for sleep homeostatic drive because of the corre- lation of this measure with the difference in cumulative time awake versus asleep [44, 50, 51]. However, as Davis et al [65] point out, some pharmacological manipulations can dissociate alterations in delta power and changes in vigilance states, thus confounding NREM delta power as a metric of sleep homeostatic drive. This situation is clearly applicable to the current dataset, where CB1 antagonists increased delta power across all vigilance states without chang- ing total sleep time. On the contrary, CB1 antagonists fragment NREM sleep, and this may lead one to speculate that the augmentation of delta is a response to poor sleep quality, and therefore, an increase in sleep drive. In fact, sleep fragmentation protocols are associated with increasing homeostatic sleep drive [66–68]. However, evidence from the current dataset argues

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 31 / 47 Endocannabinoid Signaling Regulates Sleep Stability

against the hypothesis that the CB1 antagonist-induced increase in delta power is indicative of an increase in sleep homeostatic processes. First, the time course for augmented delta does not reflect a gradual increase that one would expect if homeostatic sleep drive were building over time with reduced sleep quality. Second, the rebound in total NREM sleep following 6 Hr sleep deprivation was not occluded by CB1 antagonists. Third, blockade of CB1 did not alter the rate of recovery from sleep deprivation. It should be noted that we did not normalize our power spectral data as in other reports [17, 28, 29, 32], and our presentation of raw power spectral results may explain some discrepancies between our findings and those of others. It is not clear why other researchers have not observed the large augmentation of delta and theta power following administration of CB1 antagonists reported here. In fact, several reports have observed reduced delta and theta power following administration of CB1 antagonists [19, 28] while others have concluded that these drugs have lit- tle or no effect on the EEG power spectrum [29, 32]. We can only speculate that the variability in methods used to normalize power spectral bandwidths has contributed to these inconsistencies. Although we observed no changes in low-frequency oscillations following drugs that activate eCB signaling (JZL, AM3506, CP47), gamma oscillations were consistently reduced, particularly during NREM and REM sleep. This result is consistent with previous reports suggesting reduced gamma power is a robust effect of increased CB1 activation [69–73]. Another consistent effect of cannabinoid agonists is to facilitate high-voltage spindles/spike wave discharges (HVS) [74–76], but in the present study, power spectral analysis found no changes in the HVS bandwidth. This is not surprising given that spindles are rare in C57 mice under baseline conditions [77, 78], so at 3–12 Hr time bins used here, spindle events would only account for a small fraction of the NREM epochs. Thus, it is highly likely that any changes would be averaged out with the analysis performed here. In all cases where cannabinoids have been reported to augment spindles and their respective power spectral correlates [74–76], some prior detection method has been imple- mented to isolate epochs containing these events before comparing power spectral features across drug conditions. Consequently, no conclusions regarding cannabinoid-induced changes in the incidence of spindles can be derived from the present results.

eCB Signaling and Sleep Previous reports of CB1 antagonist effects in animal studies of sleep indicated either subtle aug- mentation of wake at the expense of NREM [15, 19, 28, 29] or no effect [13, 31, 32]. The subtle differences observed with respect to CB1 antagonist effects on sleep time across these studies are likely due to different doses and times of administration. Additionally, most of these studies used relatively short recordings (4–8 Hr). In the reports indicating increased wake time follow- ing administration of CB1 antagonists, the arousing properties of these drugs were only observed when summating across the entire recording. On the other hand, fragmentation of NREM has not been reported, but only a few studies have examined metrics of sleep architec- ture following administration of CB1 antagonists [29, 32]. Studies in CB1 KO mice have found significantly reduced NREM bout duration with an increased number of NREM bouts [24, 25]. REM sleep is consistently reduced following administration of CB1 antagonists [28, 29, 32, 79, 80], and reduced REM sleep is frequently associated with NREM fragmentation [67, 68]. In the present study, reductions in REM sleep time were associated with time points where there was noticeable fragmentation of NREM. Thus, it is likely that NREM fragmentation is a common outcome of CB1 antagonism, and this may give rise to reduced REM. Importantly, a large reduction in REM sleep was seen following JZL administration before the LP. This finding was not as evident when this drug was administered prior to the DP, likely due to a floor effect as REM is very infrequent the DP. However, some impairment of REM was

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 32 / 47 Endocannabinoid Signaling Regulates Sleep Stability

seen in all experiments in the present study with eCB/CB1 system activation and by others with THC administration [81], suggesting that suppression of REM is a consistent effect of globally increasing eCB signaling. Nevertheless, the similarity between the effect of JZL and AM281 to reduce REM is striking. In contrast to the NREM fragmentation evoked by AM281, facilitating eCB/CB1 signaling stabilizes NREM sleep, and the similar suppression of REM seen with these two opposite manipulations indicates that an optimal level of NREM stabilization must be achieved to allow for REM to emerge. On the neurobiological level, it is possible that the REM suppressing effects of CB1 antagonism and activation arise from different circuit ele- ments controlling vigilance states. This would not be surprising for the eCB system, consider- ing its molecular constituents are widespread throughout structures known to control sleep- wake transitions and considering CB1-activation reduces neurotransmission at both excitatory and inhibitory synapses. However, all drug administrations were performed systemically, so no specific conclusions about neurobiological loci can be drawn from the results presented herein. According to the influential two-process model, sleep timing is regulated by interactions between homeostatic and circadian processes, where homeostatic control of sleep is defined as a mechanism that increases the propensity to sleep as a function of the amount of time spent awake [44, 51]. Changes in the amount of sleep arise from modulation of one of these two pro- cesses. Our data do not support a role for eCB signaling in sleep homeostasis. First, CB1 activa- tion causes a biphasic response in NREM sleep time, increasing sleep during the DP and reducing it during the LP. This reduction in sleep time during the LP (secondary response) is the expected homeostatic response to increased sleep during the DP. Second, blockade of CB1 signaling does not substantially alter sleep time. While the duration of NREM bouts is reduced by CB1 antagonism, the compensatory increase in the number of bouts argues for an intact homeostatic mechanism. Finally, there was not a robust effect of CB1 blockade on rebound sleep time or the amount of sleep recovered following TSD. Consequently, we conclude that it is unlikely that eCB signaling is an essential component of sleep homeostatic machinery.

Relevance and Future Directions In both rodents and humans, chronic cannabinoid administration produces down-regulation of CB1 signaling [82, 83], and although sleep disturbances are a central feature of cannabis withdrawal in humans [84], there is marked paucity of objective studies describing exactly how sleep is disrupted much less a mechanism for these disruptions [85]. Interestingly, insomnia and poor sleep quality were commonly reported adverse effects in clinical trials with the CB1 antagonist rimonabant, but again, no objective measures of sleep were obtained [22, 23]. The rimonabant trial and trials with other CB1 antagonists were terminated due to the increased depression, anxiety, and suicidality that was associated with these drugs, so it is unlikely that any data pertaining to how CB1 antagonists affect human sleep will be forthcoming in the near future. Nevertheless, the self-reports of sleep disturbances from these clinical trials are interest- ing considering the association between insomnia and depression [86]. Our findings demon- strate that eCB signaling is necessary and sufficient for the control of sleep stability, but this neurotransmitter system is not necessary for sleep homeostasis.

Supporting Information S1 Data. Zip File Containing Data Relevant to This Manuscript. (ZIP) S1 Fig. Sleep Deprivation Apparatus. A, Exploded schematic view of structural components of the sleep deprivation chambers labelled with dimensions in inches. B, Photograph of an

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 33 / 47 Endocannabinoid Signaling Regulates Sleep Stability

assembled device and a schematic of a top down view of the chamber. Note that the rotation described in the schematic implies rotation of the chamber floor/disc suspended beneath the clear acrylic chamber wall. Also note the commutator and tether in the photograph. Polysom- nographic activity can be recorded in these chambers during the sleep deprivation. C, Sche- matic overview of assembled system including electronic control components: break-our board (BoB), Arduino, TTL controlled relay to control power circuit to motors, voltage regulator, and computer running custom control software written in MATLAB. (PDF) S2 Fig. Vehicle Solution Does Not Alter Sleep Parameters in C57BL/6J Mice. Data are from experiment with CP47 (N = 9), where subjects were administered a saline injection i.p. the day prior to the vehicle injection. The vehicle data depicted here are the same as those depicted in Fig 4. A, NREM sleep time or architecture. Top graph: Percent time in NREM was not affected by the vehicle solution. Middle graph: The duration of NREM bouts was not affected by vehicle injection. Bottom graph: The number of NREM bouts was not affected by the vehicle solution. B, REM sleep time and architecture. Top graph: The percent time in REM was not affected by vehicle injection. Middle graph: The duration of REM bouts was not affected by vehicle injec- tion. Bottom graph: For the number of REM bouts, there was an overall interaction (treatment x time of day within photoperiod, F(6,98.77) = 2.63, p = 0.021), nested interaction (time of day within photoperiod, F(6, 95.49) = 6.56, p < 0.001), and a main effect of treatment (F(1, 74.92) = 82.37, p < 0.001). Overall, the vehicle solution did not alter the number of REM bouts when data were collapsed across the day or when comparisons were made with data col- lapsed within either LP or DP. However, there was a slight reduction in the number of REM bouts at one point in the LP (ZT06-09: t(82.02) = -2.10, p = 0.039). Given the small effect size, limited to only one measure of REM architecture in a very restricted timeframe many hours after the injection, we conclude that the vehicle solution used in this study has little or no effect on sleep in C57BL/6 mice. C-E, There were no obvious changes in EEG power spectra follow- ing vehicle administration. C, Power spectra from wake epochs. D, Power spectra from NREM epochs. E, Power spectra from REM epochs. In A & B, Grey shaded regions indicate the DP, and symbols/bars represent means±SEM across all subjects for each 3 Hr time bin. (PDF) S3 Fig. Percent Agreement Between Automated and Human Scoring of Data by Vigilance State. To compute percent agreement by vigilance state, each human’s score and the comput- er’s score were compared against a template derived from human scored data. This meant that for each human there were two possible templates, and these values were averaged together yielding one human:human percent agreement score per human scorer per each of 5 data files used (the data file with corrupt EMG channel used in overall percent agreement, Fig 1C, was excluded for this analysis as it was unscorable). Thus, for each data file there were three human:human measures and three computer:human measures. The state-specific percent agreement was calculated as the fraction of epochs where the scorer and template agreed that epochs were or were not a target state over the total number of epochs (% agreement = 100% x [agree State + agree not State]/total number of epochs). For each state (wake, NREM, and REM) a two-way repeated measures ANOVA was performed with data file as a repeated factor and scoring comparison (human:human vs. computer:human) as a between-groups factor. A, Results for percent agreement for wake epochs. There was an interaction between scoring type and datafile (F(4,16) = 3.82, p = 0.023) and a main effect of data file (F(4,16) = 21.92, p < 0.001). However, there was a only a slight reduction in percent agreement for data file number 5 in the human:computer (t(20) = 4.14, p = 0.003). B, Results for percent agreement for NREM epochs. There was only a main effect of data file (F(4, 16) = 28.53, p < 0.001). C,

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 34 / 47 Endocannabinoid Signaling Regulates Sleep Stability

Results for percent agreement for REM epochs. There was only a main effect of data file (F(4, 16) = 33.18, p < 0.001). D, Shows data collapsed across scorers and the results of a paired com- parison by data file. Grey, connected points superimposed on the bar graph indicate mean human:human and computer:human agreement for each data file. A paired t-test was per- formed for each vigilance state. Percent agreement was not significantly different for human: human vs human:computer scored data for wake (t(4) = 1.94, p = 0.12), NREM (t(4) = 1.09, p 0.34), or REM (t(4) = 0.77, p = 0.48). (PDF) S4 Fig. MAGL Inhibition with JZL184 Administration Before the LP Attenuates Gamma Frequency Oscillations During Sleep. A-C, Average power spectra for epochs of different vig- ilance states across the entire LP (left hand) and DP (right hand). Solid lines denote means and shaded region around lines denotes SEM. A, Wake. B, NREM. C, REM. D-F, Change over the day in summated power in different frequency bandwidths from the power spectra: delta (left hand column), theta (middle column), and gamma (right hand column). D, Wake epochs. Left panel: For wake delta, there was an overall interaction (treatment x time of day within photope- riod, F(24, 261.30) = 2.08, p = 0.003), nested interaction (time of day within photoperiod, F(6, 242.10) = 9.33, p < 0.001), and a main effect of photoperiod (F(1, 134.75) = 6.88, p = 0.010). The only time point that significantly deviated from vehicle was during the first 3 Hr of the recovery day, when there was an increase in delta power (t(215.67) = 2.86, p = 0.018). Middle panel: No effect of JZL184 on wake theta power. Right panel: For wake gamma power, there was a nested interaction (time of day within photoperiod, F(6, 253.19) = 6.08, p < 0.001) and main effects of both treatment (F(4,67.43) = 3.21, p = 0.018) and photoperiod (F(1, 179.51) = 115.90, p < 0.001). Specifically, 16 mg/kg JZL reduced gamma power during the first 3 Hr of the LP (ZT 00–03: t(57.18) = -2.68, p = 0.038). E, NREM epochs. Left panel: For NREM delta power, there was no effect of JZL treatment. Middle panel: For NREM theta power, there was an overall interaction (treatment x time of day within photoperiod, F(24, 268.23) = 1.64, p = 0.033), a nested interaction (time of day within photoperiod, F(6, 238.31) = 20.36, p < 0.001), and a main effect of photoperiod (F(1, 159.84) = 85.90, p < 0.001). However, there were no specific time points where JZL184 significantly altered NREM theta power relative to vehicle. Right panel: For NREM gamma power, there was an overall interaction (treatment x time of day within photoperiod, F(24, 267.36) = 2.46, p < 0.001), a nested interaction (time of day within photoperiod, F(6, 234.97) = 31.18, p < 0.001), and main effects of both treatment (F (4, 163.42) = 22.79, p < 0.001) and photoperiod (F(1, 126.56) = 230.95, p < 0.001). Overall, JZL had dose-dependent effects on NREM gamma, with decreases observed following 8.0 (t (145.58) = -3.45, p = 0.003) and 16.0 mg/kg doses (t(94.93) = -4.97, p < 0.001). 8.0 mg/kg JZL decreased NREM gamma power for 18 Hr following drug administration (ZT 00–18: t(179.23) -2.71, p 0.030), and 16.0 mg/kg JZL reduced NREM gamma across the entire recording (ZT 00–00: t(116.56) -2.97, p 0.015). F, REM epochs. Left panel: JZL had no effect on REM delta power. Middle panel: JZL had no effect on REM theta power. Right panel: For REM gamma, there was an overall interaction (treatment x time of day within photoperiod, F(24, 236.78) = 1.81, p = 0.014), a secondary interaction (treatment x photoperiod, F(4, 235.193) = 4.82, p = 0.001), a nested interaction (time of day within photoperiod, F(6, 217.40) = 15.30, p < 0.001), and main effects of both treatment (F(4, 96.44) = 14.58, p < 0.001) and photope- riod (F(1, 121.40) = 77.38, p < 0.001). There was an overall dose-dependent effect of JZL on REM gamma power, with reduced gamma following both 8.0 (t(80.55) = -2.97, p = 0.015) and 16.0 mg/kg doses (t(49.88) = -4.32, p < 0.001). The 8.0 mg/kg dose reduced REM gamma across all time points in the first 15 Hr following administration (ZT 00–15: t(113.90) -2.58, p 0.044), while the 16.0 mg/kg dose decreased gamma power for 18 Hr after administration

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 35 / 47 Endocannabinoid Signaling Regulates Sleep Stability

(ZT 00–18: t(67.41) -3.12, p 0.011). Symbols/Bars represent mean±SEM for 3 hr time bins (N = 8). Grey background in graphs shows dark photoperiod. Asterisks denote significant dif- ference from vehicle baseline. All injections administered at onset of LP (ZT 00:00). (PDF) S5 Fig. Direct Activation of CB1 with CP47,497 Attenuates Gamma Frequency Oscillations During Sleep. A-C, Average power spectra for epochs of different vigilance states across the entire DP (left hand) and LP (right hand). Solid lines denote means and shaded region around lines denotes SEM. A, Wake. B, NREM. C, REM. D-F, Change over the day in summated power in different frequency bandwidths from the power spectra: delta (left hand column), theta (middle column), and gamma (right hand column). D, Wake Epochs. Left panel: CP47 had no effect on wake delta power. Middle panel: For wake theta power, there was a significant overall interaction (drug x time of day within photoperiod, F(15, 180.83) = 9.73, p < 0.001) with a significant reduction in wake theta at only a single time point during the dark photope- riod (ZT 15–18: t(186.39) = -2.29, p = 0.047). Right panel: For wake gamma, there was a signifi- cant overall interaction (drug x time of day within photoperiod, F(15, 179.99) = 3.04, p < 0.001) with a main effect of photoperiod (F(1, 135.02) = 6.86, p = 0.010). However, there was not a difference at any specific time point between low or high dose CP47 and vehicle. E, NREM Epochs. Left panel: For NREM delta there was an overall interaction (drug x time of day within photoperiod, F(15, 179.22) = 3.07, p < 0.001). However, there were no pair-wise dif- ferences at any time point between high or low dose CP47 and vehicle. Middle panel: For NREM theta power, there was an overall interaction (F(15, 180.85) = 2.79, p = 0.001) with main effect of photoperiod (F(1, 157.14) = 50.99, p < 0.001). However, there were no pair-wise difference between drug treatment conditions and vehicle. Right panel: For NREM gamma, there was an overall interaction (drug x time of day within photoperiod, F(15, 181.48) = 3.50, p < 0.001), secondary interaction (drug x photoperiod, F(2, 184.76) = 8.82, p < 0.001) with main effects of drug treatment (F(2, 175.98) = 7.13, p < 0.001) and photoperiod (F(1, 174.89) = 11.39, p = 0.001). Specifically, 1.0 mg/kg CP47 reduced NREM gamma power during the first 9 Hr of the dark photoperiod (ZT12-21: t(191.79) = -2.38, p = 0.037). F, REM Epochs. Left panel: There was no effect of CP47 treatment on REM delta power. Middle panel: For REM theta power, there was an overall interaction (drug x time of day within photoperiod, F(15, 164.20) = 2.55, p = 0.002). High dose CP47 reduced REM theta power only during the first 3 Hr of recording (ZT12-15: t(150.16) = 3.14, p = 0.004). Right panel: For REM gamma, there was an overall interaction (drug x time of day within photoperiod, F(15, 167.92) = 6.02, p < 0.001) with a main effect of photoperiod (F(1, 134.36) = 14.73, p < 0.001). Specifically, high dose CP47 reduced REM gamma power during the first 6 Hr of the dark photoperiod (ZT 12–18: t(172.91) = -3.20, p 0.003). Symbols/Bars represent mean±SEM for 3 hr time bins (N = 9). Grey background in graphs shows dark photoperiod. Asterisks denote significant dif- ference from vehicle baseline. All injections administered at onset of DP (ZT 12:00). (PDF) S6 Fig. URB597 Does Not Produce Substantial Effects on Sleep When Administered Sys- temically. A, Diagram of experimental protocol for recording sleep after administration of the reversible FAAH inhibitor, URB597. B, Effect of URB597 on NREM sleep time and architec- ture. Left Graph: There was no effect of URB on NREM sleep time. Middle Graph: For NREM bout duration, there was a nested interaction (time of day within photoperiod, F(22,609.62) = 3.04, p < 0.001) and main effects of treatment (F(4, 312.20) = 56.45, p < 0.001) and photope- riod (F(1,365.61) = 80.16, p < 0.001). 10.0 mg/kg URB produced and overall increase in NREM bout duration (t(303.24) = 3.40, p = 0.003), specifically during the third hour of the DP (ZT14-15: t(1002.87) = 2.82, p = 0.020). Right Graph: For the number of NREM bouts, there

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 36 / 47 Endocannabinoid Signaling Regulates Sleep Stability

was a nested interaction (time of day within photoperiod, F(22,690.70) = 1.60, p = 0.041) and main effects of both treatment (F(4,198.00) = 2.97, p = 0.021) and photoperiod (F(1,253.69) = 88.28, p < 0.001). Overall, 10.0 mg/kg URB reduced the number of NREM bouts (t(194.74) = -2.84, p = 0.020), but there were no differences at specific time points. C, Effect of URB597 of REM sleep time and architecture. Left Graph: There was no effect of URB on REM sleep time. Middle Graph: For REM bout duration, there was a nested interaction (time of day within pho- toperiod, F(22,478.80) = 2.31, p = 0.001) and main effects of drug treatment (F(4,249.61) = 3.80, p = 0.005) and photoperiod (F(1,302.99) = 11.14, p = 0.001). Overall, 10.0 mg/kg URB increased REM bout duration, specifically during the fifth hour of the DP (ZT16-17: t(726.45) = 2.54, p = 0.045). Right Graph: For the number of REM bouts, there was a nested interaction (time of day within photoperiod, F(22,266.50) = 4.84, p = 0.001), and main effects of treatment (F(4, 266.50) = 4.84, p = 0.001) and photoperiod (F(1,318.73) = 152.30, p < 0.001). Overall, 10.0 mg/kg URB reduced the number of REM bouts (t(263.14) = -2.90, p = 0.016), but there were several specific time points throughout the day when 10.0 mg/kg URB decreased the number of REM bouts (ZT17-18, 04–05, 09–10: t(955.91) -2.54, p 0.045). Symbols/Bars represent mean±SEM for 1 Hr time bins (N = 10). Grey background in graphs shows dark pho- toperiod. Asterisks denote significant difference from vehicle baseline. (PDF) S7 Fig. Long-lasting FAAH Inhibition with AM3506 Attenuates Gamma Frequency Oscil- lations During Sleep. A-C, Average power spectra for epochs of different vigilance states across the entire DP (left hand) and LP (right hand). Solid lines denote means and shaded region around lines denotes SEM. A, Wake. B, NREM. C, REM. D-F, Change over the day in summated power in different frequency bandwidths from the power spectra: delta (left hand column), theta (middle column), and gamma (right hand column). D, Wake epochs. Left panel: For wake delta power, there was an overall interaction (drug x time of day within photo- period, F(12, 182.27) = 2.47, p = 0.005) and a secondary interaction (drug x photoperiod, F(2, 185.16) = 3.62, p = 0.029) with a main effect of photoperiod (F(1, 176.67) = 34.22, p < 0.001). Specifically, there was increased wake delta power during the first half of the dark photoperiod on the recovery day (ZT 12–18: t(191.85) 2.94, p 0.007). Middle panel: For wake theta power, there was an overall interaction (drug x time of day within photoperiod, F(12, 180.96) = 2.05, p = 0.022), but there were no pair-wise differences at any time point on either the drug or recovery days and vehicle. Right panel: No effect of AM3506 treatment on wake gamma power. E, NREM epochs. Left panel: For NREM delta power, there was an overall interaction (drug x time of day within photoperiod, F(12, 178.50) = 3.88, p < 0.001) with a main effect of photoperiod (F(1, 115.69) = 6.88, p = 0.010). On the recovery day, NREM delta was elevated at one point during the dark photoperiod (ZT 15–18: t(151.37) = 2.55, p = 0.024). Middle panel: No effect of AM3506 on NREM theta. Right panel: For NREM gamma power, there was a secondary interaction (drug x photoperiod, F(2, 185.23) = 3.54, p = 0.031) and a nested interaction (time of day within photoperiod, F(6, 180.37) = 8.35, p < 0.001) with a main effect of drug treatment (F(2, 171.77) = 9.94, p < 0.001). Overall, AM3506 reduced NREM gamma (t(155.32) = -3.45, p = 0.001), and relative to vehicle, AM3506 specifically reduced NREM gamma across the first 18 Hr of the experiment (ZT 12–06: t(170.04) -2.42, p 0.033). F, REM epochs. Left panel: No effect of AM3506 on REM delta power. Middle panel: For REM theta, there was a significant overall interaction (drug x time of day within photoperiod, F(12, 157.56) = 1.96, p = 0.031) with a main effect of photoperiod (F(1, 104.31) = 6.78, p = 0.011). Compared to vehicle, REM theta was increased during the first 3 Hr im- mediately following administration of AM3506 (t(63.88) = 3.17, p = 0.005). Right panel: For REM gamma power, there was an overall interaction (drug x time of day within photoperiod,

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 37 / 47 Endocannabinoid Signaling Regulates Sleep Stability

F(12, 166.38) = 2.09, p = 0.020), a secondary interaction (drug x photoperiod, F(2, 172.23) = 3.29, p = 0.039) with main effects of drug (F(2, 148.36) = 8.64, p < 0.001) and photoperiod (F(1, 146.71) = 19.90, p < 0.001). Overall, REM gamma power was reduced following AM3506 administration (t(125.06) = -2.67, p = 0.017) with specific reduction in REM gamma during the first 9 Hr of the experiment (ZT 12–21: t(143.63) -3.10, p 0.005). Symbols/Bars repre- sent mean±SEM for 3 Hr time bins (N = 9). Grey background in graphs shows dark photope- riod. Asterisks denote significant difference from vehicle baseline. All injections administered at onset of dark photoperiod (ZT 12:00). (PDF) S8 Fig. Power Spectral Features of EEG are Altered by Administration of the CB1 Antago- nist AM281 Before the Dark Photoperiod. A—C, Power spectra from different vigilance states averaged over 12 Hr light/dark photoperiods. Dark lines represent group means and shaded regions surrounding the lines represent SEM. A, Wake. B, NREM. C, REM. D–F, Quantification of delta (0–4 Hz), theta (4–8 Hz), and gamma (30-60Hz) bandwidths of power spectra across the three vigilance states. D, Wake epochs. Left panel: For delta power, there was a significant overall interaction (drug x time of day within photoperiod, F(12, 232.82) = 2.05, p = 0.021) with a main effect of drug treatment (F(2, 61.15) = 12.80, p < 0.001). 5.0 mg/kg AM281 significantly increased delta power during wake epochs across most of the experiment (ZT 12–21 & 00–09; t(144.52) 2.71, p 0.015). Middle panel: For wake theta power, there was a nested interaction (time of day within photoperiod, F(6, 237.52) = 3.78, p = 0.001) with a main effect of drug treatment (F(2, 72.91) = 7.53, p < 0.001). Theta power was increased at one time point during the dark photoperiod (ZT 15–18: t(159.69) = 2.18, p = 0.049) and the first half of the light photoperiod (ZT 00–06: t(159.69) 2.87, p 0.009). Right panel: There was no effect of treatment on power in the gamma bandwidth. E, NREM epochs. Left panel: For NREM delta power, there was a significant overall interaction (drug x time of day within pho- toperiod, F(12, 243.32) = 2.77, p = 0.002), a secondary interaction (drug x photoperiod, F(2, 248.67) = 6.84, p = 0.001), a nested interaction (time of day within photoperiod, F(6, 232.06) = 30.63, p < 0.001), and main effects of both drug treatment (F(2, 104.70) = 35.00, p < 0.001) and photoperiod (F(1, 156.74) = 45.51, p < 0.001). Specifically, high dose AM281 increased NREM delta power across most time bins (ZT 12–21: t(195.82) 2.89, p 0.009). Middle panel: For NREM theta power, there was a significant overall interaction (drug x time of day within photoperiod, F(12, 247.79) = 2.40, p = 0.006), a nested interaction (time of day within photoperiod, F(6, 238.8) = 11.47, p < 0.001), and main effects of both photoperiod (F(1, 187.29) = 58.41, p < 0.001) and treatment (F(2, 189.81) = 19.12, p < 0.001). Pair-wise compari- sons found that high dose AM281 increased NREM theta power across much of the day start- ing 3 Hr after drug administration (ZT 15–21: t(250.10) 2.32, p 0.042). Right panel: For NREM gamma, there was an overall interaction (F(12, 248.00) = 5.11, p < 0.001), a nested interaction (time of day within photoperiod, F(6,241.11) = 10.37, p < 0.001), and a main effect of photoperiod (F(1, 198.44) = 83.52, p < 0.001). Following AM281 administration, NREM gamma power was increased at only one time point during the dark photoperiod (ZT 06–09: t (255.10) = 2.60, p = 0.020). F, REM epochs. Left panel: For REM delta power, there was a sec- ondary interaction (drug x photoperiod, F(2, 167.31) = 7.24, p = 0.001), a nested interaction (time of day within photoperiod, F(6, 204.83) = 4.16, p = 0.001), and main effects of drug treat- ment (F(2, 61.28) = 14.18, p < 0.001) and photoperiod (F(1, 195.42) = 38.71, p < 0.001). Spe- cifically, AM281 increased REM delta across all time points in the dark photoperiod (ZT12-00: t(172.89) = 3.20, p 0.003). Middle panel: For REM theta power, there was an overall interac- tion (drug x time of day within photoperiod, F(12, 211.73) = 2.20, p = 0.013), a nested interac- tion (time of day within photoperiod, F(6, 208.14) = 2.52, p = 0.023), and a main effect of drug

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 38 / 47 Endocannabinoid Signaling Regulates Sleep Stability

treatment (F(2, 79.84) = 9.51, p < 0.001). There were several time points over the day when AM281 administration increased REM theta power (ZT 15–03: t(164.93) 2.42, p 0.033). Right panel: There was no effect of AM281 treatment on REM gamma power. Symbols/Bars represent mean±SEM for 3 hr time bins (N = 12). Grey background in graphs shows dark pho- toperiod. Asterisks denote significant difference from vehicle baseline. All injections adminis- tered at onset of DP (ZT 12:00). (PDF) S9 Fig. Blockade of CB1 Does Not Alter Recovery of Sleep Following 6 Hr TSD. To deter- mine if sleep homeostatic mechanisms were altered by administration of the CB1 antagonist immediately following 6 Hr of acute TSD, we determined the sleep deficit incurred during TSD and computed the recovery from this deficit. A, First the cumulative NREM sleep time was cal- culated for each subject in both groups over sequential 3 Hr bins across three phases of the experiment: baseline vehicle administration, sleep deprivation, and the recovery day immedi- ately following sleep deprivation. B, Second, each subject’s baseline cumulative sleep was sub- tracted from each of these three curves. The baseline day is only shown here to demonstrate this normalization. C, The sleep debt incurred by each animal after sleep deprivation was taken as the last bin of the baseline normalized cumulative sleep on the deprivation day. This value was separately calculated for each subject and was subtracted from each point of the baseline normalized cumulative sleep plot for the recovery day (open symbols panel B). This yielded the two curves depicted in panel C for the recovery from sleep debt incurred during TSD. A two- way repeated measures ANOVA was performed on the NREM recovery data shown in panel C with treatment group as a between-groups factor and time of day as a within-subjects repeated measure. There was a significant main effect of time of day (F(7, 126) = 51.62, p < 0.001), but there was neither an interaction (F(7, 126) = 1.35, p = 0.23) nor a main effect of treatment (F(1, 18) = 0.75, p = 0.40), suggesting that both groups recovered similarly from TSD. In panels A and B, the red arrows denote the two time bins when the sleep deprivation device was acti- vated. There was generally lower overall sleep for most of sleep deprivation day in both groups, even prior to activation of the rotor. However, this is not surprising given that the subjects had just been placed into the deprivation chambers and were likely habituating to the new environ- ment. Symbols/Bars represent mean±SEM for 3 Hr time bins. Grey background in graphs shows dark photoperiod. Injections were delivered on baseline and recovery days half-way through the LP (ZT 06:00). On the baseline day, both groups received a vehicle injection. On the recovery day, the vehicle group (N = 11) received another vehicle injection, while the AM281 group (N = 9) received a 5.0 mg/kg injection of AM281. (PDF) S10 Fig. Treatment with AM281 Results in a Late Rebound in REM. A, Diagram of experi- mental protocol repeated here for clarity with different color coding to indicate measures reflect REM sleep parameters. B, Overall fluctuation in REM sleep throughout the entire exper- iment. Downward facing arrows denote times at which injections were given. The red horizon- tal line indicates the time at which the sleep deprivation chambers were activated. C, Comparisons within and between treatment groups across the first baseline and recovery days. Top Graph: Percent time in REM sleep. There was a significant overall interaction (treatment group x time of day within photoperiod within experimental phase, F(24, 270) = 7.01, p < 0.001), a secondary interaction (treatment group x photoperiod within experimental phase, F(3, 270) = 3.56, p = 0.015), a tertiary interaction (treatment group x experimental phase, F(1, 270) = 6.71, p = 0.010), and a main effect of photoperiod (F(1, 270) = 32.26, p < 0.001). For the vehicle group, there was significantly less REM sleep overall on the recovery day compared to baseline (t(270) = -2.64, p = 0.009), but there was not an overall difference

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 39 / 47 Endocannabinoid Signaling Regulates Sleep Stability

between groups for the amount of REM on either the baseline or recovery days. However, the AM281 group had significantly more REM sleep than vehicle treated mice towards the end of the recovery day (ZT 21–06: t(98.63) 2.01, p 0.047). Compared to their baseline sleep, both vehicle and AM281 treated mice had significantly less REM during the first 3 Hr of the recording (ZT 06–09: t(270) -3.13, p 0.002), but the AM281 group went on to exhibit a REM rebound late in the recording (ZT 21–06: t(270) 2.12, p 0.035), while the vehicle group continued to have less REM than their baseline (ZT 00–03: t(270) = -2.51, p = 0.013). Middle Graph: REM bout duration. There was no effect of either sleep deprivation or AM281 treatment on REM bout duration in this experiment. This is possibly because the estimates of REM bout duration were taken from a small number of REM bouts for each subject following sleep deprivation. Bottom graph: For the number of REM bouts, there was an overall interac- tion (treatment group x time of day within photoperiod within experimental phase, F(24, 270) = 6.78, p < 0.001), a secondary interaction (treatment group x photoperiod within experimental phase, F(3, 270) = 9.94, p < 0.001), a tertiary interaction (treatment group x experimental phase, F(1, 270) = 4.20, p = 0.041), and a main effect of photoperiod (F(1, 270) = 11.22, p = 0.001). However, the AM281 group had fewer REM bouts than vehicle group during the DP on the baseline day (t(40.83) = -2.86, p = 0.007). Specifically, on the baseline day the AM281 group had fewer REM bouts than the vehicle group, during the first 9 Hr of the DP (ZT12-21: t(160.70) = -2.34, p = 0.020) and significantly more REM bouts during the first 3 Hr of the subsequent LP (ZT00-03, t(160.70) = 2.21, p = 0.029). Consequently, the accentuated baseline circadian fluctuation in REM in the AM281 group should be taken into account when interpreting between-group differences during recovery. Relative to their own baselines, both groups had a reduction in REM bouts during the first 3 Hr of the recovery (ZT06-09: t(270) -2.29, p = 0.023). For the vehicle group, this reduction in the number of REM bouts continued into the DP (ZT18-21: t(270) = -4.07, p < 0.001), while for the AM281 group, the number of REM bouts increased above baseline levels late in the recovery (ZT00-06: t(270) 3.14, p 0.002). Late in the recovery day, the AM281 group had significantly more REM bouts com- pared to vehicle treated mice at the same time points where the AM281 group had more REM bouts than its own baseline (ZT00-06: t(270) 4.03, p < 0.001). Additionally, considering there were baseline differences between groups at only one of these time points (ZT00-03), it is likely that this late augmentation in the number of REM bouts reflects a real phenomenon and is not a reflection of a pre-existing group difference. Nevertheless, obtaining a measure where baseline differences do not exist between groups would provide more confidence that another, uncontrolled factor is not involved in this process. Green symbols/lines show AM281 group (N = 9) while the vehicle group (N = 11) is depicted in black. Downward facing arrows (#) indi- cate time of drug administration. Asterisks () denote significant pair-wise comparisons within-groups between drug conditions and measures obtained during vehicle baseline. Dag- gers (†) denote significant pair-wise comparisons between groups on the recovery day. Pound symbols (#) denote significant pair-wise comparisons between groups on the baseline day. The legend on the bottom right corresponds to panel C.InB & C, Grey shaded regions indicate the DP, and symbols/bars represent means±SEM across all subjects for each 3 Hr time bin (PDF) S11 Fig. Blockade of CB1 During Recovery from TSD Significantly Augments Low Fre- quency Power Spectral Features, Confounding Attempts to Use These as Indices of Sleep Homeostatic Drive. A, Diagram of experimental protocol for sleep deprivation. B, Wake theta power during sleep deprivation. Horizontal red bar indicates sleep deprivation session during first 6 Hr of the LP. C, Wake epochs. Left panel: For wake delta, there was an overall interac- tion (treatment group x time of day within photoperiod within experimental phase, F(12,

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 40 / 47 Endocannabinoid Signaling Regulates Sleep Stability

259.47) = 12.27, p < 0.001) and a secondary interaction (treatment group x experimental phase, F(1, 192.08) = 16.66, p < 0.001). For the first 6 Hr following TSD, wake delta was increased by AM281 administration relative to delta power measurements in the vehicle group (ZT06-12: t(35.02) 3.20, p 0.003). Additionally, within-group comparisons found that AM281 elevated wake delta across the first 12 Hr of recovery from TSD (ZT06-18: t(255.42) 2.09, p 0.038). In contrast, wake delta power was reduced in the vehicle treated group but only during the last 3 Hr of the recovery (ZT03-06: t(255.42) = -2.04, p = 0.042). Middle panel: For wake theta, there was an overall interaction (treatment group x time of day within photope- riod within experimental phase, F(12, 259.47) = 12.27, p < 0.001) and a secondary interaction (treatment group x experimental phase, F(1, 185.67) = 6.26, p = 0.013). There were no pair- wise differences between groups during baseline or recovery. However, treatment with AM281 increased wake theta relative to baseline during the first 12 Hr of recovery (ZT06-18: t(252.32) 2.18, p 0.030). For the vehicle group, there were no pair-wise differences in wake theta between baseline and recovery. Right panel: For wake gamma, there was a nested interaction (time of day within photoperiod within experimental phase, F(12, 255.50) = 13.77, p < 0.001) and a main effect of experimental phase (F(1, 139.37) = 9.85, p = 0.002). Across treatment groups, there was an overall reduction of wake gamma power during recovery from TSD rela- tive to baseline (t(139.37) = -3.14, p = 0.002). D, NREM epochs. Left panel: For NREM delta, there was an overall interaction (treatment group x time of day within photoperiod within experimental phase, F(12, 255.42) = 8.84, p < 0.001), secondary interaction (treatment x exper- imental phase, F(1, 156.93) = 25.84, p < 0.001), and a main effect of experimental phase (F(1, 156.93) = 6.37, p = 0.013). Relative to the vehicle group, AM281 administration during recov- ery from TSD increased NREM delta power across the first 15 Hr of the recording (ZT06-21: t (38.44) 2.60, p 0.013). Within-groups comparisons between recovery and baseline found increased NREM delta power following AM281 administration after TSD across the majority of the recording (ZT06-00: t(233.13) 2.18, p 0.030). For the vehicle group, NREM delta during recovery was reduced relative to baseline measures during the first 9 Hr of the DP (ZT12-21: t(233.13) -2.27, p 0.024). Middle panel: For NREM theta power, there was an overall interaction (treatment group x time of day within photoperiod within experimental phase, F(12, 263.39) = 5.07, p < 0.001) and a secondary interaction (treatment group x experi- mental phase, F(1, 238.82) = 18.25, p < 0.001). Between groups comparisons on the recovery day found increased theta power in the AM281-treated group (ZT06-21: t(29.42) 3.04, p 0.005), but there were no differences between groups during baseline. Within groups com- parisons found elevated NREM theta in the AM281 group during recovery relative to baseline (ZT06-21: t(277.23) 2.84, p 0.005). In contrast, NREM theta was reduced at several time points during recovery in the vehicle-treated group (ZT09-21: t(277.23) -2.01, p 0.045). Right panel: For NREM gamma power, there was an overall interaction (treatment group x time of day within photoperiod within experimental phase, F(12, 264.25) = 2.73, p < 0.002). There were no pair-wise differences between groups during baseline or recovery. However, for the AM281 group, NREM gamma was increased relative to baseline for the first 6 Hr of the DP (ZT12-18: t(281.46) 2.44, p 0.015). E, REM Epochs. Left panel: For REM delta power, there was a secondary interaction between treatment group and experimental phase (F(1, 85.32) = 6.01, p = 0.016). There were no differences between groups during either baseline or recovery phases of the experiments. However, for the AM281 group, there was an overall increase in REM delta during recovery relative to baseline (t(86.08) = 2.80, p = 0.006). There was no difference between baseline and recovery for the vehicle treated group. Middle panel: For REM theta power, there was an overall interaction (treatment group x time of day within photoperiod within experimental phase, F(12, 216.66) = 1.83, p = 0.045) with main effects of treatment group (F(1, 28.36) = 18.34, p < 0.001) and experimental phase (F(1, 59.83) = 4.26,

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 41 / 47 Endocannabinoid Signaling Regulates Sleep Stability

p = 0.043). There were no differences between treatment groups during baseline, but REM theta was elevated at all time points of the recovery day relative to the vehicle group (ZT 06–06: t(106.40) 1.99, p 0.049). Within-groups comparisons between recovery and baseline found increased REM theta during the first 9 Hr of recovery in the AM281 group (ZT 06–15: t (170.25) 2.31, p 0.022). There were no differences in REM theta between recovery and baseline recordings for the vehicle treated group. Right panel: There was no effect of TSD or AM281 treatment on REM gamma power. For B-D: Grey shaded regions denote dark photope- riod. Open symbols with dotted lines indicate data from the recovery day 1 while closed sym- bols with solid lines represent data from the baseline day 1. Asterisks () denote significant pair-wise comparisons within-groups between drug conditions and measures obtained during vehicle baseline. Daggers (†) denote significant pair-wise comparisons between groups on the recovery day. Symbols/bars represent means±SEM across all subjects for each 3 Hr time bin. For AM281 group N = 9, and for the vehicle group N = 11. (PDF)

Acknowledgments This work was supported by the United States National Institutes of Health. DML and MJP received support from the National Institute on Alcohol Abuse and Alcoholism's Division of Intramural Clinical and Biological Research, award number ZIA AA000416 (http://www.niaaa. nih.gov/). AM received support from National Institute on Drug Abuse Grant DA003801 (http://www.drugabuse.gov/). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. We wish to thank Dr. Shih- Chieh Lin (NIA, NIH) for his advice on the computational data analysis used here and for his feedback on this manuscript. Additionally, we wish to thank Dr. John J. Woodward (Medical University of South Carolina, Charleston, SC) and Dr. Igor Timofeev (Université Laval, Qué- bec, Canada) for their comments during the preparation of this manuscript.

Author Contributions Conceived and designed the experiments: MJP. Performed the experiments: MJP. Analyzed the data: MJP. Contributed reagents/materials/analysis tools: AM. Wrote the paper: MJP DML.

References 1. Russo E. Cannabis in India: ancient lore and modern medicine. In: Mechoulam R, editor. Cannabinoids as Therapeutics. Milestones in drug therapy. Basel; Boston: Birkhäuser; 2005. p. 1–22. 2. O'Shaughnessy WB. On the Preparations of the Indian Hemp, or Gunjah: Cannabis Indica Their Effects on the Animal System in Health, and their Utility in the Treatment of Tetanus and other Convulsive Dis- eases. Prov Med J Retrosp Med Sci. 1843; 5(123):363–9. PubMed Central PMCID: PMCPMC2490264. 3. Wallich GC. Cannabis Indica. Br Med J. 1883; 1(1173):1224. PMID: 20750654. 4. Bradbury JB. The Croonian Lectures on some Points Connected with Sleep, Sleeplessness, and Hyp- notics: Delivered before the Royal College of Physicians of London. Br Med J. 1899; 2(2011):134–8. PMID: 20758585. 5. Clendinning J. Observations on the medicinal properties of the Cannabis Sativa of India. Medico-chirur- gical transactions. 1843; 26:188–210. PMC2116906. PMID: 20895771 6. Moreton JE, Davis WM. Electroencephalographic study of the effects of tetrahydrocannabinols on sleep in the rat. Neuropharmacology. 1973; 12(9):897–907. Epub 1973/09/01. PMID: 4355678. 7. Watanabe K, Narimatsu S, Yamamoto I, Yoshimura H. Difference in tolerance development of hypo- thermia and pentobarbital-induced sleep prolongating effect of 11-hydroxy-delta 8-tetrahydrocannabi- nol and 11-oxo-delta 8-tetrahydrocannabinol in mice. European journal of pharmacology. 1982; 77 (1):53–6. Epub 1982/01/08. PMID: 6277654.

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 42 / 47 Endocannabinoid Signaling Regulates Sleep Stability

8. Watanabe K, Narimatsu S, Yamamoto I, Yoshimura H. Cross-tolerance development to the prolonga- tion of pentobarbitone-induced sleep by delta 8-tetrahydrocannabinol and 11-hydroxy-delta 8-tetrahy- drocannabinol in mice. The Journal of pharmacy and pharmacology. 1987; 39(11):945–7. Epub 1987/ 11/01. PMID: 2892923. 9. Zarcone VP Jr. Marijuana and ethanol: effects on sleep. Int J Psychiatry Med. 1973; 4(2):201–12. PMID: 4352042. 10. Fujimori M, Himwich HE. Delta 9-tetrahydrocannabinol and the sleep-wakefulness cycle in rabbits. Physiology & behavior. 1973; 11(3):291–5. Epub 1973/09/01. PMID: 4355439. 11. Wallach MB, Gershon S. The effects of delta8-THC on the EEG, reticular multiple unit activity and sleep of cats. European journal of pharmacology. 1973; 24(2):172–8. PMID: 4358207. 12. Kano M, Ohno-Shosaku T, Hashimotodani Y, Uchigashima M, Watanabe M. Endocannabinoid-medi- ated control of synaptic transmission. Physiol Rev. 2009; 89(1):309–80. doi: 10.1152/physrev.00019. 2008 PMID: 19126760. 13. Murillo-Rodriguez E, Cabeza R, Mendez-Diaz M, Navarro L, Prospero-Garcia O. Anandamide-induced sleep is blocked by SR141716A, a CB1 receptor antagonist and by U73122, a phospholipase C inhibi- tor. Neuroreport. 2001; 12(10):2131–6. Epub 2001/07/12. PMID: 11447321. 14. Murillo-Rodriguez E, Sanchez-Alavez M, Navarro L, Martinez-Gonzalez D, Drucker-Colin R, Prospero- Garcia O. Anandamide modulates sleep and memory in rats. Brain research. 1998; 812(1–2):270–4. Epub 1998/11/14. PMID: 9813364. 15. Murillo-Rodriguez E, Blanco-Centurion C, Sanchez C, Piomelli D, Shiromani PJ. Anandamide enhances extracellular levels of adenosine and induces sleep: an in vivo microdialysis study. Sleep. 2003; 26(8):943–7. Epub 2004/01/30. PMID: 14746372. 16. Rueda-Orozco PE, Soria-Gomez E, Montes-Rodriguez CJ, Perez-Morales M, Prospero-Garcia O. Intrahippocampal administration of anandamide increases REM sleep. Neuroscience letters. 2010; 473 (2):158–62. Epub 2010/03/02. doi: 10.1016/j.neulet.2010.02.044 PMID: 20188142. 17. Murillo-Rodriguez E, Palomero-Rivero M, Millan-Aldaco D, Arias-Carrion O, Drucker-Colin R. Adminis- tration of URB597, oleoylethanolamide or palmitoylethanolamide increases waking and dopamine in rats. PloS one. 2011; 6(7):e20766. doi: 10.1371/journal.pone.0020766 PMID: 21779318; PubMed Cen- tral PMCID: PMC3136458. 18. Murillo-Rodriguez E, Vazquez E, Millan-Aldaco D, Palomero-Rivero M, Drucker-Colin R. Effects of the fatty acid amide hydrolase inhibitor URB597 on the sleep-wake cycle, c-Fos expression and dopamine levels of the rat. European journal of pharmacology. 2007; 562(1–2):82–91. Epub 2007/03/06. doi: 10. 1016/j.ejphar.2007.01.076 PMID: 17336288. 19. Murillo-Rodriguez E, Millan-Aldaco D, Di Marzo V, Drucker-Colin R. The anandamide membrane trans- porter inhibitor, VDM-11, modulates sleep and c-Fos expression in the rat brain. Neuroscience. 2008; 157(1):1–11. Epub 2008/09/30. doi: 10.1016/j.neuroscience.2008.08.056 PMID: 18822353. 20. Murillo-Rodriguez E, Palomero-Rivero M, Millan-Aldaco D, Di Marzo V. The administration of endocan- nabinoid uptake inhibitors OMDM-2 or VDM-11 promotes sleep and decreases extracellular levels of dopamine in rats. Physiology & behavior. 2013; 109:88–95. Epub 2012/12/15. doi: 10.1016/j.physbeh. 2012.11.007 PMID: 23238438. 21. Huitron-Resendiz S, Sanchez-Alavez M, Wills DN, Cravatt BF, Henriksen SJ. Characterization of the sleep-wake patterns in mice lacking fatty acid amide hydrolase. Sleep. 2004; 27(5):857–65. PMID: 15453543. 22. Nathan PJ, O'Neill BV, Napolitano A, Bullmore ET. Neuropsychiatric adverse effects of centrally acting antiobesity drugs. CNS Neurosci Ther. 2011; 17(5):490–505. doi: 10.1111/j.1755-5949.2010.00172.x PMID: 21951371. 23. Steinberg BA, Cannon CP. Cannabinoid-1 receptor blockade in cardiometabolic risk reduction: safety, tolerability, and therapeutic potential. Am J Cardiol. 2007; 100(12A):27P–32P. doi: 10.1016/j.amjcard. 2007.10.011 PMID: 18154743. 24. Pava MJ, den Hartog CR, Blanco-Centurion C, Shiromani PJ, Woodward JJ. Endocannabinoid modu- lation of cortical up-states and NREM sleep. PloS one. 2014; 9(2):e88672. Epub 2014/02/13. doi: 10. 1371/journal.pone.0088672 PMID: 24520411; PubMed Central PMCID: PMC3919802. 25. Silvani A, Berteotti C, Bastianini S, Lo Martire V, Mazza R, Pagotto U, et al. Multiple sleep alterations in mice lacking cannabinoid type 1 receptors. PloS one. 2014; 9(2):e89432. Epub 2014/03/04. doi: 10. 1371/journal.pone.0089432 PMID: 24586776; PubMed Central PMCID: PMC3930731. 26. Zimmer A, Zimmer AM, Hohmann AG, Herkenham M, Bonner TI. Increased mortality, hypoactivity, and hypoalgesia in cannabinoid CB1 receptor knockout mice. Proceedings of the National Academy of Sci- ences of the United States of America. 1999; 96(10):5780–5. PMID: 10318961; PubMed Central PMCID: PMCPMC21937.

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 43 / 47 Endocannabinoid Signaling Regulates Sleep Stability

27. Wu CS, Zhu J, Wager-Miller J, Wang S, O'Leary D, Monory K, et al. Requirement of cannabinoid CB(1) receptors in cortical pyramidal neurons for appropriate development of corticothalamic and thalamocor- tical projections. The European journal of neuroscience. 2010; 32(5):693–706. doi: 10.1111/j.1460- 9568.2010.07337.x PMID: 21050275; PubMed Central PMCID: PMCPMC2970673. 28. Santucci V, Storme JJ, Soubrie P, Le Fur G. Arousal-enhancing properties of the CB1 cannabinoid receptor antagonist SR 141716A in rats as assessed by electroencephalographic spectral and sleep- waking cycle analysis. Life sciences. 1996; 58(6):PL103–10. Epub 1996/01/01. PMID: 8569415. 29. Goonawardena AV, Plano A, Robinson L, Ross R, Greig I, Pertwee RG, et al. Modulation of food con- sumption and sleep-wake cycle in mice by the neutral CB1 antagonist ABD459. Behav Pharmacol. 2015; 26(3):289–303. doi: 10.1097/FBP.0000000000000108 PMID: 25356730; PubMed Central PMCID: PMC4445652. 30. Perez-Morales M, Alvarado-Capuleno I, Lopez-Colome AM, Mendez-Diaz M, Ruiz-Contreras AE, Pros- pero-Garcia O. Activation of PAR1 in the lateral hypothalamus of rats enhances food intake and REMS through CB1R. Neuroreport. 2012; 23(14):814–8. Epub 2012/08/15. doi: 10.1097/WNR. 0b013e328357615a PMID: 22889888. 31. Mendelson WB, Basile AS. The hypnotic actions of oleamide are blocked by a cannabinoid receptor antagonist. Neuroreport. 1999; 10(15):3237–9. PMID: 10574567. 32. Goonawardena AV, Plano A, Robinson L, Platt B, Hampson RE, Riedel G. A Pilot Study into the Effects of the CB1 Cannabinoid Receptor Agonist WIN55,212–2 or the Antagonist/Inverse Agonist AM251 on Sleep in Rats. Sleep disorders. 2011; 2011:178469. Epub 2011/01/01. doi: 10.1155/2011/178469 PMID: 23471192; PubMed Central PMCID: PMC3581240. 33. Liedhegner ES, Sasman A, Hillard CJ. Brain region-specific changes in N-acylethanolamine contents with time of day. J Neurochem. 2014; 128(4):491–506. doi: 10.1111/jnc.12495 PMID: 24138639; PubMed Central PMCID: PMC3946166. 34. Valenti M, Vigano D, Casico MG, Rubino T, Steardo L, Parolaro D, et al. Differential diurnal variations of anandamide and 2-arachidonoyl-glycerol levels in rat brain. Cell Mol Life Sci. 2004; 61(7–8):945–50. doi: 10.1007/s00018-003-3453-5 PMID: 15095014. 35. National Research Council Committee for the Update of the Guide for the Care and Use of Laboratory Animals, Institute for Laboratory Animal Research, National Academies Press. Guide for the care and use of laboratory animals. 8th ed. Washington, D.C.: National Academies Press; 2011. xxv, 220 p. p. 36. Gervasoni D, Lin SC, Ribeiro S, Soares ES, Pantoja J, Nicolelis MA. Global forebrain dynamics predict rat behavioral states and their transitions. The Journal of neuroscience: the official journal of the Society for Neuroscience. 2004; 24(49):11137–47. Epub 2004/12/14. doi: 10.1523/JNEUROSCI.3524-04.2004 PMID: 15590930. 37. Mang GM, Franken P. Sleep and EEG Phenotyping in Mice. Curr Protoc Mouse Biol. 2012; 2(1):55–74. doi: 10.1002/9780470942390.mo110126 PMID: 26069005. 38. Diniz Behn CG, Klerman EB, Mochizuki T, Lin SC, Scammell TE. Abnormal sleep/wake dynamics in orexin knockout mice. Sleep. 2010; 33(3):297–306. PMID: 20337187; PubMed Central PMCID: PMCPMC2831423. 39. Bastianini S, Berteotti C, Gabrielli A, Del Vecchio F, Amici R, Alexandre C, et al. SCOPRISM: a new algorithm for automatic sleep scoring in mice. J Neurosci Methods. 2014; 235:277–84. doi: 10.1016/j. jneumeth.2014.07.018 PMID: 25092499. 40. Weber F, Chung S, Beier KT, Xu M, Luo L, Dan Y. Control of REM sleep by ventral medulla GABAergic neurons. Nature. 2015; 526(7573):435–8. doi: 10.1038/nature14979 PMID: 26444238. 41. Srinivasan BV, Qi H, Duraiswami R. GPUML: Graphical processors for speeding up kernel machines. Siam Conference on Data Mining; April 2010; Columbus, OH2010. 42. Vaughn LK, Denning G, Stuhr KL, de Wit H, Hill MN, Hillard CJ. Endocannabinoid signalling: has it got rhythm? British journal of pharmacology. 2010; 160(3):530–43. Epub 2010/07/02. doi: 10.1111/j.1476- 5381.2010.00790.x PMID: 20590563; PubMed Central PMCID: PMC2931554. 43. Vyazovskiy VV, Riedner BA, Cirelli C, Tononi G. Sleep homeostasis and cortical synchronization: II. A local field potential study of sleep slow waves in the rat. Sleep. 2007; 30(12):1631–42. PMID: 18246973; PubMed Central PMCID: PMCPMC2276140. 44. Borbely AA, Achermann P. Sleep homeostasis and models of sleep regulation. J Biol Rhythms. 1999; 14(6):557–68. PMID: 10643753. 45. Kim T, Ramesh V, Dworak M, Choi DS, McCarley RW, Kalinchuk AV, et al. Disrupted sleep-wake regu- lation in type 1 equilibrative nucleoside transporter knockout mice. Neuroscience. 2015; 303:211–9. doi: 10.1016/j.neuroscience.2015.06.037 PMID: 26143012; PubMed Central PMCID: PMCPMC4532636.

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 44 / 47 Endocannabinoid Signaling Regulates Sleep Stability

46. Jacobs J. Hippocampal theta oscillations are slower in humans than in rodents: implications for models of spatial navigation and memory. Philos Trans R Soc Lond B Biol Sci. 2014; 369(1635):20130304. doi: 10.1098/rstb.2013.0304 PMID: 24366145; PubMed Central PMCID: PMCPMC3866455. 47. Fries P. Neuronal gamma-band synchronization as a fundamental process in cortical computation. Annu Rev Neurosci. 2009; 32:209–24. doi: 10.1146/annurev.neuro.051508.135603 PMID: 19400723. 48. Fegley D, Gaetani S, Duranti A, Tontini A, Mor M, Tarzia G, et al. Characterization of the fatty acid amide hydrolase inhibitor cyclohexyl carbamic acid 3'-carbamoyl-biphenyl-3-yl ester (URB597): effects on anandamide and oleoylethanolamide deactivation. J Pharmacol Exp Ther. 2005; 313(1):352–8. doi: 10.1124/jpet.104.078980 PMID: 15579492. 49. Gunduz-Cinar O, MacPherson KP, Cinar R, Gamble-George J, Sugden K, Williams B, et al. Conver- gent translational evidence of a role for anandamide in amygdala-mediated fear extinction, threat pro- cessing and stress-reactivity. Mol Psychiatry. 2013; 18(7):813–23. doi: 10.1038/mp.2012.72 PMID: 22688188; PubMed Central PMCID: PMC3549323. 50. Dijk DJ, Brunner DP, Beersma DG, Borbely AA. Electroencephalogram power density and slow wave sleep as a function of prior waking and circadian phase. Sleep. 1990; 13(5):430–40. PMID: 2287855. 51. Borbely AA. A two process model of sleep regulation. Hum Neurobiol. 1982; 1(3):195–204. PMID: 7185792. 52. Dzirasa K, Ribeiro S, Costa R, Santos LM, Lin SC, Grosmark A, et al. Dopaminergic control of sleep- wake states. The Journal of neuroscience: the official journal of the Society for Neuroscience. 2006; 26 (41):10577–89. doi: 10.1523/JNEUROSCI.1767-06.2006 PMID: 17035544. 53. Lin SC, Gervasoni D. Defining Global Brain States Using Multielectrode Field Potential Recordings. In: Nicolelis MAL, editor. Methods for Neural Ensemble Recordings. Frontiers in Neuroscience. 2nd ed. Boca Raton (FL)2008. 54. Sunagawa GA, Sei H, Shimba S, Urade Y, Ueda HR. FASTER: an unsupervised fully automated sleep staging method for mice. Genes to cells: devoted to molecular & cellular mechanisms. 2013; 18 (6):502–18. doi: 10.1111/gtc.12053 PMID: 23621645; PubMed Central PMCID: PMC3712478. 55. Savinainen JR, Jarvinen T, Laine K, Laitinen JT. Despite substantial degradation, 2-arachidonoylgly- cerol is a potent full efficacy agonist mediating CB(1) receptor-dependent G-protein activation in rat cer- ebellar membranes. British journal of pharmacology. 2001; 134(3):664–72. doi: 10.1038/sj.bjp. 0704297 PMID: 11588122; PubMed Central PMCID: PMCPMC1572991. 56. Melvin LS, Milne GM, Johnson MR, Subramaniam B, Wilken GH, Howlett AC. Structure-activity rela- tionships for cannabinoid receptor-binding and analgesic activity: studies of bicyclic cannabinoid ana- logs. Mol Pharmacol. 1993; 44(5):1008–15. PMID: 8246904. 57. Compton DR, Johnson MR, Melvin LS, Martin BR. Pharmacological profile of a series of bicyclic canna- binoid analogs: classification as cannabimimetic agents. J Pharmacol Exp Ther. 1992; 260(1):201–9. PMID: 1309872. 58. Mackie K, Devane WA, Hille B. Anandamide, an endogenous cannabinoid, inhibits calcium currents as a partial agonist in N18 neuroblastoma cells. Mol Pharmacol. 1993; 44(3):498–503. PMID: 8371711. 59. Garzon J, de la Torre-Madrid E, Rodriguez-Munoz M, Vicente-Sanchez A, Sanchez-Blazquez P. Gz mediates the long-lasting desensitization of brain CB1 receptors and is essential for cross-tolerance with morphine. Mol Pain. 2009; 5:11. doi: 10.1186/1744-8069-5-11 PMID: 19284549; PubMed Central PMCID: PMC2657119. 60. Zhuang S, Kittler J, Grigorenko EV, Kirby MT, Sim LJ, Hampson RE, et al. Effects of long-term expo- sure to delta9-THC on expression of cannabinoid receptor (CB1) mRNA in different rat brain regions. Brain Res Mol Brain Res. 1998; 62(2):141–9. PMID: 9813289. 61. Martinez-Vargas M, Morales-Gomez J, Gonzalez-Rivera R, Hernandez-Enriquez C, Perez-Arredondo A, Estrada-Rojo F, et al. Does the neuroprotective role of anandamide display diurnal variations? Int J Mol Sci. 2013; 14(12):23341–55. doi: 10.3390/ijms141223341 PMID: 24287910; PubMed Central PMCID: PMCPMC3876049. 62. Yasuo S, Koch M, Schmidt H, Ziebell S, Bojunga J, Geisslinger G, et al. An endocannabinoid system is localized to the hypophysial pars tuberalis of Syrian hamsters and responds to photoperiodic changes. Cell Tissue Res. 2010; 340(1):127–36. doi: 10.1007/s00441-010-0930-7 PMID: 20165884. 63. Acuna-Goycolea C, Obrietan K, van den Pol AN. Cannabinoids excite circadian clock neurons. The Journal of neuroscience: the official journal of the Society for Neuroscience. 2010; 30(30):10061–6. doi: 10.1523/JNEUROSCI.5838-09.2010 PMID: 20668190; PubMed Central PMCID: PMCPMC2927117. 64. Sanford AE, Castillo E, Gannon RL. Cannabinoids and hamster circadian activity rhythms. Brain research. 2008; 1222:141–8. doi: 10.1016/j.brainres.2008.05.048 PMID: 18582849.

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 45 / 47 Endocannabinoid Signaling Regulates Sleep Stability

65. Davis CJ, Clinton JM, Jewett KA, Zielinski MR, Krueger JM. Delta wave power: an independent sleep phenotype or epiphenomenon? J Clin Sleep Med. 2011; 7(5 Suppl):S16–8. doi: 10.5664/JCSM.1346 PMID: 22003323; PubMed Central PMCID: PMC3190419. 66. Baud MO, Magistretti PJ, Petit JM. Sustained sleep fragmentation induces sleep homeostasis in mice. Sleep. 2015; 38(4):567–79. doi: 10.5665/sleep.4572 PMID: 25325477; PubMed Central PMCID: PMCPMC4355896. 67. He J, Kastin AJ, Wang Y, Pan W. Sleep fragmentation has differential effects on obese and lean mice. J Mol Neurosci. 2015; 55(3):644–52. doi: 10.1007/s12031-014-0403-7 PMID: 25152064; PubMed Cen- tral PMCID: PMC4320048. 68. Ringgold KM, Barf RP, George A, Sutton BC, Opp MR. Prolonged sleep fragmentation of mice exacer- bates febrile responses to lipopolysaccharide. J Neurosci Methods. 2013; 219(1):104–12. doi: 10.1016/ j.jneumeth.2013.07.008 PMID: 23872243; PubMed Central PMCID: PMC3993011. 69. Hajos M, Hoffmann WE, Kocsis B. Activation of cannabinoid-1 receptors disrupts sensory gating and neuronal oscillation: relevance to schizophrenia. Biol Psychiatry. 2008; 63(11):1075–83. doi: 10.1016/j. biopsych.2007.12.005 PMID: 18261715. 70. Hajos N, Katona I, Naiem SS, MacKie K, Ledent C, Mody I, et al. Cannabinoids inhibit hippocampal GABAergic transmission and network oscillations. The European journal of neuroscience. 2000; 12 (9):3239–49. PMID: 10998107. 71. Kucewicz MT, Tricklebank MD, Bogacz R, Jones MW. Dysfunctional prefrontal cortical network activity and interactions following cannabinoid receptor activation. The Journal of neuroscience: the official journal of the Society for Neuroscience. 2011; 31(43):15560–8. doi: 10.1523/JNEUROSCI.2970-11. 2011 PMID: 22031901. 72. Cortes-Briones J, Skosnik PD, Mathalon D, Cahill J, Pittman B, Williams A, et al. Delta(9)-THC Disrupts Gamma (gamma)-Band Neural Oscillations in Humans. Neuropsychopharmacology: official publication of the American College of Neuropsychopharmacology. 2015; 40(9):2124–34. doi: 10.1038/npp.2015. 53 PMID: 25709097. 73. Robbe D, Montgomery SM, Thome A, Rueda-Orozco PE, McNaughton BL, Buzsaki G. Cannabinoids reveal importance of spike timing coordination in hippocampal function. Nat Neurosci. 2006; 9 (12):1526–33. doi: 10.1038/nn1801 PMID: 17115043. 74. Buonamici M, Young GA, Khazan N. Effects of acute delta 9-THC administration on EEG and EEG power spectra in the rat. Neuropharmacology. 1982; 21(8):825–9. PMID: 6289162. 75. Sales-Carbonell C, Rueda-Orozco PE, Soria-Gomez E, Buzsaki G, Marsicano G, Robbe D. Striatal GABAergic and cortical glutamatergic neurons mediate contrasting effects of cannabinoids on cortical network synchrony. Proceedings of the National Academy of Sciences of the United States of America. 2013; 110(2):719–24. Epub 2012/12/28. doi: 10.1073/pnas.1217144110 PMID: 23269835; PubMed Central PMCID: PMC3545808. 76. Turkanis SA, Karler R. Central excitatory properties of delta 9-tetrahydrocannabinol and its metabolites in iron-induced epileptic rats. Neuropharmacology. 1982; 21(1):7–13. PMID: 6278353. 77. Kim D, Hwang E, Lee M, Sung H, Choi JH. Characterization of topographically specific sleep spindles in mice. Sleep. 2015; 38(1):85–96. doi: 10.5665/sleep.4330 PMID: 25325451; PubMed Central PMCID: PMCPMC4262960. 78. Ryan LJ. Characterization of cortical spindles in DBA/2 and C57BL/6 inbred mice. Brain Res Bull. 1984; 13(4):549–58. PMID: 6441615. 79. Perez-Morales M, De La Herran-Arita AK, Mendez-Diaz M, Ruiz-Contreras AE, Drucker-Colin R, Pros- pero-Garcia O. 2-AG into the lateral hypothalamus increases REM sleep and cFos expression in mela- nin concentrating hormone neurons in rats. Pharmacology, biochemistry, and behavior. 2013; 108:1–7. Epub 2013/04/23. doi: 10.1016/j.pbb.2013.04.006 PMID: 23603032. 80. Jacobson LH, Commerford SR, Gerber SP, Chen YA, Dardik B, Chaperon F, et al. Characterization of a novel, brain-penetrating CB1 receptor inverse agonist: metabolic profile in diet-induced obese models and aspects of central activity. Naunyn-Schmiedeberg's archives of pharmacology. 2011; 384(6):565– 81. Epub 2011/09/29. doi: 10.1007/s00210-011-0686-y PMID: 21947251. 81. Feinberg I, Jones R, Walker J, Cavness C, Floyd T. Effects of marijuana extract and tetrahydrocannabi- nol on electroencephalographic sleep patterns. Clin Pharmacol Ther. 1976; 19(6):782–94. PMID: 178475. 82. Ceccarini J, Kuepper R, Kemels D, van Os J, Henquet C, Van Laere K. [18F]MK-9470 PET measure- ment of cannabinoid CB1 receptor availability in chronic cannabis users. Addict Biol. 2015; 20(2):357– 67. doi: 10.1111/adb.12116 PMID: 24373053. 83. Sim LJ, Hampson RE, Deadwyler SA, Childers SR. Effects of chronic treatment with delta9-tetrahydro- cannabinol on cannabinoid-stimulated [35S]GTPgammaS autoradiography in rat brain. The Journal of

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 46 / 47 Endocannabinoid Signaling Regulates Sleep Stability

neuroscience: the official journal of the Society for Neuroscience. 1996; 16(24):8057–66. PMID: 8987831. 84. American Psychiatric Association. Diagnostic and statistical manual of mental disorders: DSM-5. 5th ed. Washington, D.C.: American Psychiatric Association; 2013. xliv, 947 p. p. 85. Gates P, Albertella L, Copeland J. Cannabis withdrawal and sleep: A systematic review of human stud- ies. Subst Abus. 2015: 0. doi: 10.1080/08897077.2015.1023484 PMID: 25893849. 86. Fernandez-Mendoza J, Shea S, Vgontzas AN, Calhoun SL, Liao D, Bixler EO. Insomnia and incident depression: role of objective sleep duration and natural history. J Sleep Res. 2015; 24(4):390–8. doi: 10.1111/jsr.12285 PMID: 25728794.

PLOS ONE | DOI:10.1371/journal.pone.0152473 March 31, 2016 47 / 47 Therapeutic Benefits of Cannabis: A Patient Survey

Charles W. Webb MD and Sandra M. Webb RN, BSN

Abstract A universal pain scale was used to assess pain before and Clinical research regarding the therapeutic benefits of cannabis (“marijuana”) has been almost non-existent in the United States since cannabis was given after treatment (0 = no pain, 10 = worst pain ever). Open-ended Schedule I status in the Controlled Substances Act of 1970. In order to discover questions were asked to ascertain the following: the benefits and adverse effects perceived by medical cannabis patients, especially with regards to chronic pain, we hand-delivered surveys to one (1) “Any adverse effects you have had from using medical hundred consecutive patients who were returning for yearly re-certification cannabis?” for medical cannabis use in Hawai‘i. (2) “Does medical cannabis help you with any other The response rate was 94%. Mean and median ages were 49.3 and problems? If so, what?” 51 years respectively. Ninety-seven per cent of respondents used cannabis primarily for chronic pain. Average pain improvement on a 0-10 pain scale was 5.0 (from 7.8 to 2.8), which translates to a 64% relative decrease in average The purpose of the last question was to explore benefits out- pain. Half of all respondents also noted relief from stress/anxiety, and nearly side the parameters of the state of Hawai‘i’s medical cannabis half (45%) reported relief from insomnia. Most patients (71%) reported no qualifying conditions. adverse effects, while 6% reported a cough or throat irritation and 5% feared arrest even though medical cannabis is legal in Hawai‘i. No serious adverse Results effects were reported. The overall response rate was 94%. The mean age was 49.3 These results suggest that Cannabis is an extremely safe and effective medication for many chronic pain patients. Cannabis appears to alleviate years and the median age was 51. No data was collected on sex pain, insomnia, and may be helpful in relieving anxiety. Cannabis has shown or race/ethnicity. Almost all respondents (97%) used medical extreme promise in the treatment of numerous medical problems and deserves cannabis primarily for relief of chronic pain. to be released from the current Schedule I federal prohibition against research Average reported pain relief from medical cannabis was and prescription. substantial. Average pre-treatment pain on a zero to ten scale was 7.8, whereas average post-treatment pain was 2.8, giving Introduction a reported average improvement of 5 points. This translates to Research into the therapeutic benefits of cannabis has been a 64% average relative decrease in pain. severely limited by the federal Schedule I classification, which Other reported therapeutic benefits included relief from essentially prohibits any ability to acquire or to provide cannabis stress/anxiety (50% of respondents), relief of insomnia (45%), for studies investigating possible therapeutic effects. Limited improved appetite (12%), decreased nausea (10%), increased studies have been done in Canada and in Europe, as well as focus/concentration (9%), and relief from depression (7%). several in California. Several patients wrote notes (see below) relating that cannabis Hawai‘i is one of twenty states (plus the District of Colum- helped them to decrease or discontinue medications for pain, bia) which allow certifications for use of medical cannabis. anxiety, and insomnia. Other reported benefits did not extend The authors have been certifying patients for use of medical to 5% or more of respondents. cannabis in Hawai‘i for more than four years. In an attempt to Six patients (6%) wrote brief notes relating how cannabis discover the perceived benefits and adverse effects of medical helped them to decrease or to discontinue other medications. cannabis, we conducted a survey of medical cannabis patients. Comments included the following: “Medical cannabis replaced my need for oxycodone. Now I don’t need them at all.” “I do Methods not need Xanax anymore.” “In the last two years I have been Sample Selection able to drop meds for anxiety, sleep, and depression.” “I’ve cut Between July of 2010 and February of 2011, we hand-delivered back 18 pills on my morphine dosage.” questionnaires to one hundred consecutive patients who had A majority (71%) reported no adverse effects, while 6% been certified for the medical use of cannabis for a minimum reported a cough and/or throat irritation and 5% reported a of one year and were currently re-applying for certification. fear of arrest. All other adverse effects were less than 5%. No serious adverse effects were reported. Survey Design and Administration The subjects were verbally instructed to complete the Discussion questionnaire in the office at the time of re-certification or According to the Institute of Medicine, chronic pain afflicts 116 were provided a stamped and addressed envelope so they million Americans and costs the nation over $600 billion every could complete the questionnaire at home. All patients were year in medical treatment and lost productivity.1 Chronic pain is instructed to remain anonymous and to answer the questions a devastating disease that frequently leads to major depression as honestly as possible. and even suicide.2 Unfortunately, the therapeutic options for chronic pain are limited and extremely risky.

HAWAI‘I JOURNAL OF MEDICINE & PUBLIC HEALTH, APRIL 2014, VOL 73, NO 4 109 Spurred by efforts to encourage physicians to become more recalcitrant chronic pain. A University of Toronto systematic pro-active in treating chronic pain, US prescription opioids review of randomized controlled trials (RCT’s) examining (synthetic derivatives of opium) have increased ten-fold since cannabinoids in the treatment of chronic pain found that fifteen 1990.3 By 2009 prescription opioids were responsible for almost of eighteen trials demonstrated significant analgesic effect of half a million emergency department visits per year.4 In 2010 cannabinoids and that there were no serious adverse effects.15 prescription opioid overdoses were responsible for well over While opioids are generally considered to have little benefit 16,000 deaths.5 A 2010 article in the New England Journal of in chronic neuropathic pain, several RCT’s have shown that Medicine addressing this problem is aptly titled “A Flood of cannabinoids can relieve general neuropathic pain,16 as well as Opioids, a Rising Tide of Deaths.”3 Drugs such as OxyContinR neuropathic pain associated with HIV and with multiple scle- are so dangerous that the manufacturer’s boxed warning states rosis (MS). 17,18 One study found that cannabis had continuing that “respiratory depression, including fatal cases, may occur efficacy at the same dose for at least two years. 19 with use of OxyContin, even when the drug has been used Even low dose inhaled cannabis has been proven to reduce as recommended and not misused or abused.”6 Clearly safer neuropathic pain. In a randomized, double-blind, placebo- analgesics are needed. controlled crossover trial involving patients with refractory The Hippocratic Oath reminds to “first, do no harm.” It can- neuropathic pain, Ware, et al, found that therapeutic blood levels not be over-emphasized that there has never been a death from of THC (mean 45 ng/ml achieved by a single inhalation three overdose attributed to cannabis.7 In fact, no deaths whatsoever times a day) were much lower than those necessary to produce have been attributed to the direct effects of cannabis.7 Canna- a cannabis euphoria or “high”(> 100 ng/ml). 19 bis has a safety record that is vastly superior to all other pain Cannabis is relatively non-addicting, and patients who stop medications. using it (eg, while traveling) report no withdrawal symptoms. Many physicians worry that cannabis smoke might be as dan- One author (Webb C.) worked for 26 years in a high volume gerous as cigarette smoke; however, epidemiologic studies have emergency department where he never witnessed a single visit found no increase in oropharyngeal or pulmonary malignancies for cannabis withdrawal symptoms, whereas dramatic symptoms attributable to marijuana.8-10 Still, since smoke is something from alcohol, benzodiazepine, and/or opioid withdrawal were best avoided, medical cannabis patients are encouraged to use a daily occurrence. smokeless vaporizers which can be purchased on-line or at local So why is cannabis still held hostage by the DEA as a Schedule “smoke-shops.” In states that (unlike Hawai‘i) allow cannabis I substance? On June 18, 2010, the Hawai‘i Medical Association dispensaries, patients can purchase “vapor pens,” analogous passed a resolution stating in part that: to e-cigarettes and fully labeled regarding doses of THC and “Whereas, 1) Cannabis has little or no known withdrawal syndrome other relevant cannabinoids. and is therefore considered to be minimally or non-addicting; and Tests have proven that smoke-free vaporizers deliver THC as well or even more efficiently than smoking, and that most Whereas, 2) Cannabis has many well-known medical benefits patients prefer vaporizers over smoking.11 Like smoking, vapor- (including efficacy for anorexia, nausea, vomiting, pain, muscle spasms, and glaucoma) and is currently recommended by thou- izers allow patients to slowly titrate their medicine just to effect, sands of physicians; and analogous to IV patient-controlled analgesia (PCA) that has been so successful in hospital-based pain control. This avoids Whereas 3) Cannabis has been used by millions of people for the unwanted psychoactive side-effects often associated with many centuries with no history of recorded fatalities and with no oral medication such as prescription MarinolR (100% THC in lethal dosage ever discovered; and oil) capsules which tend to be slowly and erratically absorbed Whereas, Cannabis therefore fulfills none of the required three and are often either ineffectually weak or overpoweringly criteria (all of which are required) to maintain its current restric- strong.12,13 Because inhaled cannabis is rapid, reliable, and tion as a Schedule I substance… titratable, most patients strongly prefer inhaled cannabis over MarinolR capsules.14 The Hawai‘i Medical Association recommends that Medical While the relative safety of cannabis as medication is easily Cannabis be re-scheduled to a status that is either equal to or established, the degree of efficacy is still being established. The less restrictive than the Schedule III status of synthetic THC reported pain relief by patients in this survey is enormous. One (MarinolR), so as to reduce barriers to needed research and to reason for this is that patients were already self-selected for suc- humanely increase availability of cannabinoid medications to cess: they had already tried cannabis and found that it worked patients who may benefit.”20 for them. For this sample, the benefits of cannabis outweighed Medical cannabis remains controversial mainly because the any negative effects. The study design may therefore lend itself federal government refuses to recognize cannabis as an ac- to over-estimating the benefits and under-estimating the nega- cepted medication. To this we would echo the words of Melanie tive side-effects if extrapolated to the general population. Thernstrom in her excellent book The Pain Chronicles,2 “How Another reason that the reported pain relief is so significant could treating pain be controversial?” one might ask, “ Why is that cannabis has been proven effective for many forms of wouldn’t it be treated? Who are the opponents of relief?”

HAWAI‘I JOURNAL OF MEDICINE & PUBLIC HEALTH, APRIL 2014, VOL 73, NO 4 110 Conclusions Authors’ Affiliation: - Keauhou Urgent Care Center, 78-6831 Alii Dr., Suite 418, Kailua Kona, HI 96740 Cannabis is an extremely safe and effective medication for many patients with chronic pain. In stark contrast to opioids Correspondence to: and other available pain medications, cannabis is relatively Charles W. Webb MD; 73-993 Ahikawa St, Kailua Kona, HI 96740; non-addicting and has the best safety record of any known pain Email: [email protected] medication (no deaths attributed to overdose or direct effects References of medication). Adverse reactions are mild and can be avoided 1. Institute of Medicine of the National Academy. Relieving Pain in America. 2011. by titration of dosage using smokeless vaporizers. 2. Thernstrom M. The Pain Chronicles. New York: Farrar, Straus and Giroux; 2011. 3. Oakie S. A Flood of Opioids, a Rising Tide of Death. NEJM. 2010;363:1981-1985. More research needs to be pursued to discover degrees of 4. Substance Abuse and Mental Health Services Admin. Drug Abuse Warning Network: selected efficacy in other areas of promise such as in treating anxiety, tables of national estimates of drug-related emergency visits. Rockville, MD: Center for Behavioral Health Statistics and Quality, SAMHSA; 2010. depression, bipolar disorder, autism, nausea, vomiting, muscle 5. CDC. Opioids Drive Continued Increase in Drug Overdose Deaths. Press release Feb 20, spasms, seizures, and many neurologic disorders. Patients 2013. 6. Purdue Pharma LP. An Overview of the Oxycontin Label Update Deterrence Studies. 07/13. deserve to have cannabis released from its current federal 7. Iverson LL. The Science of Marijuana. New York: Oxford University Press; 2000. prohibition so that scientific research can proceed and so that 8. Sidney S, Beck JE, Tekawa IS, Quesenberry CP, Friedman GC. Marijuana use and mortalilty.. Am J Public Health. 1997; 87(4):585-590. physicians can prescribe cannabis with the same freedom ac- 9. Hashibe M, Straif K, Tashkin DP, Morgenstern H, Greenland S, Zhang ZF. Epidemiologic review corded any other safe and effective medications. of marijuana use and cancer risk. Alcohol. 2005;35(3):265-275. 10. Tashkin DP. Smoked marijuana as a cause of lung injury. Monaldi. Arch Chest Dis. 2005;63(2):93- 100. Conflict of Interest 11. Abrams D, et al. Vaporization as a Smokeless Cannabis Delivery System. Clin. Pharmacol. Ther. 2007;82(5):572-578. None of the authors identify a conflict of interest. 12. nstitute of Medicine. Marijuana and Medicine: Assessing the Science Base. 1999. 13. Weil A. San Francisco Chronicle. June 6, 2002. Authors’ Biography: 14. Grinspoon L. “I have yet to examine a patient who has used both smoked marijuana and Marinol Dr. Webb graduated from Dartmouth Medical School (BS Medicine) and from UC San who finds the latter more useful.” International Journal of Drug Policy. 2001 Issue. Francisco School of Medicine (MD 1974). General Residency US Public Health Hospital 15. Lynch ME, Campbell F. Cannabinoids for treatment of chronic non-cancer pain; a systemic review of randomized trials. Br, J. Clin Pharmacol. 2011 Nov; 72(5):735-44. (San Francisco) and Highland Hospital (Oakland). Emergency Medicine Physician 16. Wilsey B, et al. A randomized, placebo-controlled, crossover trial of cannabis cigarettes in 1975-2006 (Colorado), Urgent Care Physician 2007-present (Kailua Kona). Sandra neuropathic pain. J Pain. 2008;9(6):506-521. Webb RN, since 1979 (emergency and radiology nurse). Dr. Webb and nurse Webb 17. Abrams D, et al. Cannabis in Painful HIV-associated sensory neuropathy: a randomized have been certifying patients for medical use of cannabis since 2009. placebo-controlled trial. Neurology. 2007; 68(7):515-521. 18. Rog, et al. Oromucosal THC/cannabidiol for neuropathic pain associated with MS. Clin Ther. 2007;29(9):2068-2079. 19. Ware MA, Ducruet T, Robinson AR. Evaluation of herbal cannabis characteristics by medical users: a randomized trial. Harm Reduction J. 2006 Nov 13;3:32. 20. Hawai‘i Medical Association Resolution, June 18, 2010.

HAWAI‘I JOURNAL OF MEDICINE & PUBLIC HEALTH, APRIL 2014, VOL 73, NO 4 111 Copyright of Hawaii Journal of Medicine & Public Health is the property of University Clinical Education Research Associates and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. Journal of Psychoactive Drugs, 43 (2), 128-135, 2011 Copyright © Taylor & Francis Group, LLC IJ Routledge ISSN: 0279-1072 print/2i59-9777 online g^^ Taylor&Francis Croup DOI: 10.1080/02791072.2011.587700

Who Are Medical Marijuana Patients? Population Characteristics from Nine Califomia Assessment Clinics^

Craig Reinarman, Ph.D.*; Helen Nunberg, M.D., M.P.H.**; Fran Lanthier, M.A.*** & Tom Heddleston, M.A.***

Abstract — Marijuana is a currently iilegal that many physicians believe has sub- stantial therapeutic uses. The medical literature contains a growing number of studies on cannabinoids as well as case studies and anecdotal reports suggesting therapeutic potential. Fifteen states have passed medical marijuana laws, but little is known about the growing population of patients who use mari- juana medicinally. This article reports on a sample of 1,746 patients from a network of nine medical marijuana evaluation clinics in Califomia. Patienu completed a standardized medical history form; evaluating physicians completed standardized evaluation forms. From this data we describe patient characteristics, self-reported presenting symptoms, physician evaluations, other treatments tried, other drug use, and medical marijuana use practices. Pain, insomnia, and anxiety were the most common conditions for which evaluating physicians recommended medical marijuana. Shifts in the medical marijuana patient population over time, the need for further research, and the issue of diversion are discussed.

Keywords — anxiety, cannabis therapeutics, insomnia, medical marijuana, pain

Medicinal preparations containing marijuana prescribed for therapeutic use in American medical prac- (cannabis) were widely used in many societies for tice for a variety of conditions from the mid-nineteenth centuries. Dr. William O'Shaughnessy introduced it as century into the twentieth. Marijuana was admitted to a modern medicine in Europe in 1839, Marijuana was the United States Pharmacopoeia in 1850 and listed in the National Formulary and the US Dispensatory. Major authors (hank the medical marijuana patient-applicants for pharmaceutical companies including Lilly, Burroughs- providing the data, the RAND Corporation for funding data collec- Wellcome, and Parke-Davis produced cannabis-based tion and data set construction, MediCann for administrative support, the therapeutic agents (Brecher et al. 1972). Rosenbaum Foundation for financial support for this research, and Lester Grinspoon and anonymous referees for helpful comments. An earlier ver- In 1936, the Federal Bureau of Narcotics advocated a sion of this article was presented at the 59th Annual Meeting of the Society law prohibiting its use, which Congress passed in 1937, for the Study of Social Problems, San Francisco, August 9,2009. against the advice of the American Medical Association •Professor and Chair, Depanment of Sociology, University of Califomia, Santa Cruz. (Grinspoon & Bakalar 1993:9-11). This law, along with '•Private practice, Santa Cruz, CA. increased prescribing of aspirin and barbiturates, pushed •••Instructors and PhD candidates. Department of Sociology, cannabis out of the United States Pharmacopoeia and University of California, Santa Cruz. Please address correspondence and reprint requests to Craig common medical practice by 1942. Reinarman, Sociology Department, University of Califomia, Santa Cruz, After nonmedical cannabis use spread in the 1960s, CA 95064; phone: (831) 459-2617, fax: (831) 459-3518, email: craigr® the number of Americans reporting lifetime prevalence ucsc.edu

Journal of Psychoactive Drugs 128 Volume 43 (2), April - June 2011 Reinarman et ai. Wito Are Medicai Marijuana Patients? increased sharply. Recent estimates from the National challenged this policy and the U.S, Court of Appeals ruled Survey on Drug Use and Health show that 102,404,000 (in Conant v, Walters) in 2002 that it unconstitutionally Americans have used this drug, 41 % of the population aged infringed physicians' First Amendment rights to freedom 12 and over, or about half the adult population (SAMHSA of speech with their patients (McCarthy 2004), Subsequent 2010). This widespread use led to a gradual rediscovery legislation and case law have left medical marijuana (MM) of the therapeutic uses of cannabis, albeit largely without patients and their physicians in legal limbo: physician involvement. • In 2003, the California legislature passed SB 420 Alongside the spread of nonmedical use, in 1964 sci- to provide specific implementation guidelines for entists determined the precise chemical structure of delta-9 Proposition 215, including how counties should han- tetrahydrocannabinol (THC), thought to be the most sig- dle MM patient ID cards, nificant psychoactive ingredient in cannabis (Gaoni & • Most drug law enforcement is done by local Mechoulam 1964), This stimulated research in the clin- police who enforce state, not federal, drug laws. ical pharmacology of cannabinoids. Many physicians in In 2005, The California Attorney General ruled clinical practice also recognized the therapeutic potential that Proposition 215 is the legitimate will of the of cannabis (Irvine 2006; Charuvastra, Freidmann & Stein voters and is therefore valid under the California 2005), specifically, for example, for pain (Woolridge et al. Constitution for purposes of state law enforcement. 2005), as an antiemetic for chemotherapy patients (Doblin He advised the Highway Patrol and other state law & Kleiman 1991), or for symptoms of AIDS (Abrams enforcement agencies that under California law MM et al. 2003). More recently a broader medical litera- patients were legally entitled to possess and use ture documenting the therapeutic properties of endogenous cannabis for therapeutic purposes (Hoge 2005), cannabinoids has developed (e.g., NicoU & Alger 2004; • In 2006, Bush administration Attorney General Lehmann et al, 2002; Hall, Degenhart & Currow 2001), Gonzales sought to invalidate state MM laws, and Numerous case reports in the medical literature also have the U,S, Supreme Court ruled {Gonzales v. Raich suggested that cannabis has therapeutic potential for a vari- 2006) that the Compassionate Use Act—its legiti- ety of conditions. But rigorous experimental research that mate electoral provenance notwithstanding—neither might determine more precisely the therapeutic efficacy supersedes nor invalidates federal laws that prohibit of cannabis for specific conditions has been blocked by marijuana use (see Mikos 2009 for a legal analysis of the Drug Enforcement Administration (see Zeese 1999; the states' neglected power to legalize behavior that Alliance for Cannabis Therapeutics v. Drug Enforcement is criminalized under federal law), Administration 1994). • In 2008 the Supreme Court denied without comment This combination of increasing therapeutic use and an appeal by two California counties that had refused federal government opposition ultimately led to passage of to implement Proposition 215 {County of San Diego new state laws providing for the medical use of cannabis V, San Diego NORML 2008), thereby letting stand a upon physician recommendation. Since 1996, 15 U,S. lower court ruling that upheld SB 42O's provisions states and the District of Columbia have passed such regarding counties issuing MM identification cards. laws: California, Alaska, Oregon, Washington, Nevada, • In 2009, Attorney General Eric Holder issued a pol- Colorado, Maine, Montana, Michigan, and Washington, icy stating that federal drug control agencies would DC by ballot initiative; Rhode Island, New Mexico, no longer raid MM dispensaries if they operated Vermont, Hawaii, and New Jersey by state legislation. within state and local laws (Moore 2009). The first of these laws was California's Proposition • That policy notwithstanding, the DEA has continued 215, the Compassionate Use Act, passed in 1996 {San to raid MM dispensaries in California into 2011 (e,g,, Francisco Chronicle 1996). This act made it legal under Blankstein 2009). state law for patients to possess and use cannabis if recom- Within this grey area between conflicting state and mended by their physicians. Numerous medical and scien- federal laws, the number of patients who have received rec- tific associations endorsed medical use of cannabis and/or ommendations for medical marijuana from physicians has supported further research into its therapeutic poten- continued to grow, albeit by how much remains unknown. tial. These included the American College of Physicians Over 1,000 MM dispensaries, delivery services, and coop- (2008), the American Public Health Association (1995), the eratives are said to be operating in California to meet British Medical Association (1997), the Canadian Medical the demand (NORML 2007), A rough estimate of the Association (2005), and the Institute of Medicine of the number of MM patients in California can be extrapolated National Academy of Sciences (1999). from Oregon figures. Unlike California's Compassionate Such elections and endorsements notwithstanding, the Use Act, Oregon's MM law set up an Oregon Medical Bush Administration's Office of National Drug Control Marijuana Program that requires centralized record keep- Policy threatened to revoke the licenses of physicians ing. As of July, 2009, some 2,983 Oregon-licensed physi- who recommended cannabis to patients. One physician cians had approved 20,307 applications for MM (Oregon

Journal of Psychoactive Drugs i 29 Volume 43 (2), April - June 2011 Keinarman et al. Who Are Medical Maryuana Patients?

Department of Human Services 2008). The population of conventional and alternative medical treatments tried, drug California is 9.7 times that of Oregon (U.S. Census 2007), use history, and MM use practices; and (2) a physician eval- which yields a crude estimate of 196,978 MM patients uation form using International Classification of Diseases in California. This is likely an underestimate because the codes (ICD-9), Each patient received and signed an exten- California statute affords greater latitude to physicians sive informed consent form noting confidentiality, which regarding the conditions for which they can recommend was approved by the clinics' IRB. MM (", . . any other illness for which marijuana provides Most prior studies of MM patients are based on small, relief"). Americans for Safe Access (2008), a MM patient symptom-specific samples. Initially, the population of MM advocacy group, has estimated that there are well over patients in the San Francisco Bay Area were people with 200,000 physician-sanctioned MM patients in California. HIV/AIDS and cancer (e.g., Harris, Mendelson & Jones Despite their growing numbers, however, the ambigu- 1998), Later, physicians began to recommend cannabis to ous legal status of MM patients renders them a half-hidden patients with chronic pain, mood disorders and other psy- population whose characteristics are not well documented, chiatric conditions (Gieringer 2002). The data reported with the partial exception of the San Francisco Bay Area here describe what is among the largest and most symp- (O'Connell & Bou-Matar 2007; Reiman 2007a). Medical tomatically and demographically diverse samples of medi- marijuana will likely continue to be a contentious issue, cal cannabis patients to date (cf., O'Connell & Bou-Matar but across fifteen states and the District of Columbia several 2007). hundred thousand people are using marijuana as a medicine recommended by physicians, and yet little is known about RESULTS them as a patient population. We intend this study as a modest contribution toward As Table 1 indicates, the MM patients are three-fourths filling this gap. It presents data on the demographic char- male and three-fifths White. Compared to the US Census acteristics, presenting symptoms, physician evaluations, of California, the patients in this sample are on average conventional treatments tried, and MM use practices of somewhat younger, report slightly more years of formal patients from a network of MM assessment clinics in education, and are more often employed. The comparison California, also indicates that women. Latinos, and Asian Americans are underrepresented. Given the limitations of our data, we METHODS can offer only informed speculation as to why. The underrepresentation of women may be in part These data were drawn from 1,746 consecutive an epidemiological artifact of the gender distribution of admissions to nine MM assessment clinics operating in certain kinds of injuries (e.g., workplace, sports, and motor- California in July, August, and September 2006. These cycle accidents). It may also have to do with the double assessment clinics are not dispensaries and are not con- stigma women face in seeking MM—for using an illicit nected to dispensaries. They were located throughout the drug and for violating gender-specific norms against ille- state—in the north and south, coast and central valley, gal behavior in general. Moreover, as with alcohol use, and large and small cities: Modesto, Oakland, Sacramento, pregnant women and women considering pregnancy are Hollywood, San Diego, Santa Cruz, Ukiah, San Francisco, likely to have health concerns and many may fear that MM and Santa Rosa. They charged $100 to $125 for an assess- could put them in jeopardy if discovered by child protection ment. At the time our sample was drawn, these assessment agencies. clinics had evaluated over 54,000 MM patients. Without Given the high poverty rate among Latinos and their a comprehensive patient database or representative house- concentration in the manual labor end of the occupational hold surveys, there is no way to determine precisely how structure. Latinos are exposed to equal or greater risks representative this sample is of the overall population of of work-related injuries and to no less epidemiologic risk MM patients. Moreover, there is a large albeit unknown of other conditions for which MM is sometimes used. It number of people who use marijuana medicinally but who seems likely that their under-representation has to do with have not sought physician recommendations or official the undocumented status of many Latinos in California. patient ID cards, perhaps because of the expense of the The undocumented often avoid contact with government assessment. ' agencies for fear of apprehension by law enforcement, Evaluating physicians interviewed potential patients for beyond arrest and incarceration this carries the risk of and evaluated their patient medical histories for purposes of deportation. Such fears reduce the likelihood of Latinos recommending MM and issuing patient identification cards accessing health care in general and MM in particular. under the Compassionate Use Act and SB 420, The eval- Asian Americans are also underrepresented, but this may uation instruments were (1) a basic patient-administered be because they have lower prevalence of marijuana use medical history questionnaire covering demographics, pre- than other racial/ethnic groups and/or because they have senting symptoms or conditions, brief medical history. their own venerable traditions of herbal medicine.

Journal of Psychoactive Drugs. 130 Volume 43 (2), April - June 2011 Reinarman et al. Who Are Medical Marijuana Patients?

TABLE 1 TABLE 2 Demographic Characteristics of California Medical Patient Self-Reports of Therapeutic Benefits from Marijuana Patients Compared to California Medicinal Marijuana* Census 2000, Age 18 and Over {n = 1746} Percent MM U.S. Census To Relieve: Patients 2000-California Pain 82.6 Female 27.1% 50.7% Muscle Spasms 41.1 Male 72.9% 49.3% Headaches 40.7 White 61.5% 59.5% Anxiety .•^7.8 Latino 14.4% 32.4% Nausea/Vomiting 27.7 African American 11.8% 6.7% Depression 26.1 Native American 4.5% 1.0% Cramps 19.0 Asian/Pacific Islander 4.2% 11.2% Panic Attacks 16.9 Other 4.3% Diarrhea 5.0 18-24 Years Old 17.9% ~17.1% Itching 2.8 25-34 " 27.5% 15.4% To Improve: 35-44" 21.3% 16.2% Sleep 70.7 45-54 " 20.4% 12.8% Relaxation 55.1 55>" 12.6% 18.4% Appetite 37.7 23.8% * Medication Side Effects 22.5 Employed 64.8% 57.5% Anger 22.4 Health Insurance 73.4% * Involuntary Movements 6.2 Seizures 3.2 'Data not available in California Census. As Substitute for: Prescription Medication 50.9 Alcohol 13.0

African-Americans, conversely, are over-represented *N = 1,745; patients could report more than one benefit in more than in this sample. This does not appear to stem from their one category. prevalence of marijuana use, for representative national surveys show that Blacks generally do not have signif- icantly higher prevalence of marijuana use than Whites significantly smaller proportion of the total (e.g., "to relieve (SAMHSA 2005). African-Americans may be more likely nausea/vomiting" 27.7%, "to improve appetite" 37.7%) to seek MM for any of several reasons: because they and that the MM patient population has become more are disproportionately poor, more often lack health insur- diverse since the Compassionate Use Act was passed in ance, are significantly less likely to be prescribed other 1996 (cf. Ware, Adams & Guy 2005, on MM use in the medication for pain (Pletcher et al. 2008) or to receive UK, and Grotenherman 2002 on MM use in Germany). treatment for cancer (Gross et al. 2008), and because Instead, relief of pain, muscle spasms, headache, and African-Americans are a growing proportion of HIV/AIDS anxiety, as well as to improve sleep and relaxation were cases. Some of these same reasons may help to explain why the most comtnon reasons patients cited for using MM. Native Americans are also ovcrrcpresented, although their Chronic pain also topped the list of maladies for which MM proportion of both this sample and the general population was used in another California clinical sample (Reiman is too small to judge representativeness accurately. 2007b). In their medical history questionnaires, patients were Table 3 shows the ICD-9 diagnostic codes most fre- asked "Which of the following best describe the therapeu- quently recorded by evaluating physicians. Pain from back tic benefit you receive from medicinal cannabis? (Check and neck injuries was the most frequently coded. This the most important)." Patients typically reported more than appears consistent with a nationally representative Medical one therapeutic benefit (mean — 3). Early studies showed Expenditure Panel Survey, which found a 19.3% increase most patients used MM to relieve symptoms of HIV/AIDS in the prevalence of spine problems between 1997 and (Woolridge et al. 2005) or cancer, and it is likely that the 2005 (Martin et al. 2008). Back and neck pain was fol- majority of patients in our sample who reported "nausea" lowed in frequency by sleep disorders (also increasing), were cancer patients receiving chemotherapy. However, anxiety/depression, muscle spasms, and arthritis. Fully Table 2 suggests that cancer and AIDS patients are now a half of this sample reported using MM as a substitute

Journal of Psychoactive Drugs 131 Volume 43 (2), April - June 2011 Reinarman et al. Who Are Medicai Maryuana Patients?

TABLE 3 TABLE 5 Conditions Most Frequentlyr Recorded by Medical Marijuana Patients' Self-Reported Physicians As Reasons for Approving Medical Current Nonmedical Drug Use, Compared to 2006 Marijuana Patient identification Cards* National Survey on Drug Use And Health (SAMHSA 2007) Percent ICD-9 Codes Back/Spine/Neck Pain 30.6% [722.1-724.2] MM Patients NSDUH' Sleep Disorders 15.7% [307.42, 327.0] Tobacco 29.4% 25.0% Anxiety/Depression 13.0% [300.0,311.0] Alcohol 47.5 61.9 Muscle Spasms 9.5% [728.85] Cocaine 0.3 1.9 Arthritis 8.5% [715.0,721.2,721.2] Methamphetatnine 0.4 0.5 Injuries (Knee, Ankle, Foot) 4.5% [959.7] Heroin 0.1 0.3 Joint Disease/Disorders 4.4% [716.1-719.49] Other Opiates 1.2 ** Narcolepsy 3.7% [347.0] Nausea 3.4% [787.02] Note: Participants were asked "Do you currently use . . ."; answers Inflamtnation (Spine, Nerve) 2.9% [724.4] are percent responding "yes." N = 1745; patients could report more than one drug. Of smokers, 65.5% used ten or less cigarettes/day; of Headaches/Migraines 2.7% [784.0, 346.0, 346.2] drinkers, 58.7% used

TABLE 4 use was somewhat higher than in the general popula- Other TVeatment Modalities Tried for the Medical tion, but prevalence of alcohol use was significantly lower. Condition(s) for Which Patients Seek Medical Many patients reported that they valued MM because it Marijuana* allowed them to reduce their alcohol use. It is possible that self-reports on a self-administered instrument will under- % N estimate illicit drug use, particularly if patients felt that Prescription Medication 79.3% 1383 admitting illicit drug use could reduce their chances of Physical Therapy 48.7 850 obtaining a MM identification card. Rigorous assessments Chiropractic 36.3 633 of the reliability of such data must await further research, Surgery 22.3 389 but limitations aside, these data suggest low prevalence of Counseling 21.0 366 other illicit drug use among MM patients. While it is true Acupuncture 19.4 338 that the great majority of our respondents had used mari- Therapeutic Injection 15.4 269 juana recreationally, in response to a separate question over Homeopathy 12.0 209 two-fifths (41.2%) reported that they had not been using it Other Types of Treatment 11.9 208 recreationally prior to trying it for medicinal purposes. •N = 1746; patients could report multiple other treatments. Table 6 presents data on patients' medical marijuana use practices. Amoutits used per week varied from three grams or less (40.1%) to seven or more grams (23.3%). for prescription drugs, consistent with other studies (e.g., Two-thirds (67%) reported using MM daily while one- Reiman 2007a). fourth (26%) reported using less than once a week. Half Table 4 indicates that the MM patients in the sample (52.9%) reported using one or two times per day while one had tried a variety of other treatments, conventional and in ten (10%) reported using three or more times per day. alternative, for the conditions for which they were seek- Patients consumed MM primarily in the evenings (52.3%) ing a MM identification card. Four in five (79.3%) reported or prior to sleep (56.1%). More than two in five (42.3%) having tried other medications prescribed by their physi- reported that when they used depended on their medi- cians (almost half were opiates); about half (48.7%) had cal symptoms. Patients ingested MM predominantly by tried physical therapy; over a third (36.3%) had tried chi- smoking (86.1%), although one-fourth (24.4%) reported ropractic; nearly one-fourth (22.3%) reported having had ingesting orally and nearly a fourth (21.8%) reported using surgery for their condition. a vaporizer. These latter figures suggest that at least some of Table 5 compares patient responses to the drug use the time, many MM patients are choosing modes of inges- questions to those in the 2006 National Survey on Drug tion that reduce the perceived risk of harms from smoking Use and Health (SAMHSA 2007). Prevalence of tobacco (Tan et al. 2009; Hashibe et al. 2006).

Journal of Psychoactive Drugs 132 Volume 43 (2), April - June 2011 Reinarman et ai. Who Are Medical Maryuana Patients?

Further Research TABLE 6 Like other medicines, marijuana's therapeutic efficacy Medical Marijuana Use Practices varies across conditions and patient groups. This variation seems more likely when supplies remain illicit because Frequency of Medical Marijuana Use (N = 1583)^ standardized dosages or other quality controls are more dif- Daily 67.0% (1065) ficult to achieve. To gain maximum therapeutic potential 3 Times Per Day 10.0% (284) Time Of Day Typically Used (N = 1745) With regard to shifts in the patient population, it also would Prior To Sleep 56.1% (979) be very useful to have follow-up studies of patients access- Evenings 52.3% (913) ing the assessment clinics in our sample and others drawn Depends on Symptoms 42.3% (739) from similar assessment clinics. Mornings 25.7% (448) Afternoons 20.1% (350) Diversion After Work 12.4% (217) Critics have argued that some MM patients are "gam- Middle of the Night 6.5% (114) ing the system" to get marijuana for nonmedical use. All Day 5.3% (93) Neither our data nor any other data we are aware of allow Mode of Ingestion (N = 1745) any clear-cut, empirical estimate of the scale of such diver- Smoke 86.1% (1503) Oral Ingestion 24.4% (426) sion. Given the widespread nonmedical use marijuana in Vapor 21.8% (380) Ihe general population (102,404,000 Americans report life- Topical 2.8% (49) time prevalence; see SAMHSA 2010) and the risk of arrest Amount Used per Week (N = 1431) (847,864 Americans were arrested for marijuana offenses 0-3 Grams 40.1% (574) in 2008, 754,224 or 88.96% of them for possession alone; 4—7 Grams 36.5% (523) EBI 2009), it seems likely that at least some MM patients >7 Grams 23.3% (334) use MM dispensaries as sources of supply for nonmedical •Total n = 1745, but N's vary across questions because patients could use. choose more than one response and because not all responded to each Defining and measuring such diversion, however, is question. complicated at best. Given the high prevalence of nonmed- ical use, it is not surprising that most MM patients in our sample reported having used it recreationally before using it therapeutically. But as noted above, two-fifths had not been using marijuana recreationally prior to trying it for DISCUSSION medicinal purposes. Their self-reported rales of other illicit drug use are slightly lower than those found among the gen- Rediscovery of Medicinal Utility and Diversifying eral population, and their levels of educational attainment Patient Population and rate of employment are comparable to the California Compared to earlier studies of MM patients, these data population. Our data have clear limitations, but they con- suggest that the patient population has evolved from mostly tain no obvious signs that MM patients differ from the HIV/AIDS and cancer patients to a significantly more general population. diverse array. The diffusion of marijuana as a medicine Nor is drug diversion unique to medical marijuana. may have been slower than that of other medicines in con- A significant albeit unknown proportion of other patients ventional clinical practice because the flow of information obtain prescriptions for numerous drugs through legal from physician to patient is impeded by MM's ambiguous medical channels that they then use for nonmedical pur- legal status. Thus, information about the potential thera- poses, for example, Valium and other benzodiazepines peutic utility of cannabis is spread mostly via word of (Haafkens 1997), Ritalin and other stimulants prescribed mouth and other informal means. This suggests that the for ADHD, and Oxycontin and other opiates prescribed patient population is likely to continue evolving as new for pain. patients and physicians discover the therapeutic uses of The diversion issue will likely become more impor- cannabis. Ironically, this trend toward increasing thera- tant as the line between medical and nonmedical drug peutic uses is bringing marijuana back to the position it use is increasingly blurred (Murray, Gaylin & Macklin held in the U.S. Pharmacopeia prior to its prohibition in 1984). Beyond the spread of MM, Prozac and other SSRI- 1937, type antidepressants, for example, are often prescribed

Journal of Psychoactive Drugs 133 Volume 43 (2), April - June 2011 Reinarman et ai. Wiio Are Medicai Marijuana Patients?

for patients who do not meet DSM criteria for clinical patients' purposes shift to suit situational exigencies in depression but who simply feel better when taking it. Such their health and their daily lives. It is not clear where a "cosmetic psychopharmacology" (Kramer 1993) is likely border line between medical and nonmedical marijuana or to grow as new psychiatric medications come to market. other drug use might be drawn nor how it might be effec- The line between medical and nonmedical drug use has tively policed (see Reinarman & Levine 1997: 334^4), also been blurred by performance enhancing drugs such as steroids, so-called "smart drugs" that combine vitamins NOTE with psychoactive ingredients, and herbal remedies like ma huang (ephedra) available in health food stores (Burros & 1, We are grateful to one anonymous reviewer for Jay 1996). pointing out that the cost of these assessments may well These examples suggest that despite the best intentions have prevented some potential MM patients—including of physicians and law makers, much drug use does not fit many impoverished HIV/AIDS patients—from obtaining into two neat boxes, medical and nonmedical, but rather ID cards, which may have affected the demographics of this exists on a continuum where one shades into the other as sample.

REFERENCES

Abrams, D.; Hilton, J.; Leiser, R.; Shade, S. & Elbeik, T. 2003. Short- Gieringer, D. 2002. Medicai use of cannabis: Experience in California, in: term effects of cannabinoids in patients with HIV-infection. Annals F. Grotenhermen & E. Russo (Eds.) Cannabis and Cannabinoids: of Internal Medicine 139 (4): 258-66. Pharmacology, Toxicology, and Therapeutic Potential. Binghamton, Aiiiancefor Cannabis Therapeutics v. Drug Enforcement Admin. 1994. Í5 NY: Haworth. F.3d 1131 D.C. Circuit. Grinspoon, L. & Bakalar, J.B. 1993. Marijuana: The Forbidden Medicine. American College of Physicians. 2008. Supporting Research into the New Haven, CT: Yale University Press. Therapeutic Roie of Marijuana: A Position Paper. Available at Gonzales v. Raich. 2006. 352 F. 3d 1222 (Supreme Court 03-1454, http://www.acponline.org/ad vocacy/where_we_stand/other_issues/ 2006). medmarijuana.pdf Gross, C.R; Smith, B.D.; Wolf, E. & Andersen M. 2008. Racial disparities American Public Heaith Association. 1995. Access to Therapeutic in cancer therapy. Cancer 112 (4): 900-08. Marijuana/Cannabis. Available at http.7/www.apha.org/advocacy/ Grotenherman, F. 2002. The medical use of cannabis in Germany. Joumal policy/policysearch/default.htm?id=108. of Drug Issues 32: 607-34. Americans for Safe Access. 2008. Attorney General Guidelines Grotenhermen & E. Russo (Eds.) Cannabis and Cannabinoids: Signal Victory for California Campaign. http://www.safeaccessnow. Pharmacology, Toxicology, and Therapeutic Potential. Binghamton, org/articie.php?id=5562. NY: The Haworth Press. Blankstein, A. 2009. DEA, FBi, IRS raid two westside pot dispensaries: Haafkens, J. 1997. Rituais of Silence: Long-term Tranquilizer Use by Officers shoot a pit bull. Los Angeles Times, Aug. i2. Avaiiable Women in the Netherlands. Amsterdam: Het Spinhuis: Studies in at http://latimesblogs.latimes.com/lanow/2009/08/dea-fbi-irs-raids- Medicai Anthropology and Sociology. two-westside-pot-dispensaries-shoots-a-dog.html Hall, W.D.; Degenhardt, L.J. & Currow, D. 2001. Allowing the medical Brecher, E.M. and the Editors of Consumer Reports. 1972. Licit and Illicit use of cannabis. Medical Journal of Australia Í75(i): 39-40. Drugs. Boston: Little Brown. Harris, D.; Mendelson, J. E. & Jones, R.T. Í998. A survey of 100 medical British Medical Association. 1997. Therapeutic Uses of Cannabis, marijuana club members [Abstract]. In: L.S. Harris (Ed.) Problems London: Harwood. of Drug Dependence, 1998: Proceedings of the 60th Annual Canadian Medical Association. 2005. Medical Uses of Marijuana, Scientific Meeting, the College on Problems of Drug Dependence, Available at http://www.cma.ca/index.cfm/ci_id/3396/la_id.htm. Inc. NiDA Research Monograph Series, #Í79. Rockville, MD: U.S. Burros, M. & Jay, S. 1996. Concem is growing over an herb that promises Dept of Health and Human Services. a legal high. New York Times Apr. 10: Bi. Hashibe, M.; Morgenstern, H.; Cui, Y; Tashicin, D.R; Zhang, Z; Cozen, Charuvastra, A.; Freidmann, PD. & Stein, M.D. 2005. Physician atti- W.; Mack, T.M. & Greenland, S. 2006. Marijuana use and the risk of tudes regarding the prescription of medicai marijuana. Journal of iung and upper aerodigestive tract cancers: Results of a population- Addictive Diseases 24 (3): 87-94. based case-control study. Cancer Epidemiology, Biomarkers, and Conant v. Walters. 2002. 309 F. 3d. 629 (Ninth Circuit 2002, cert, denied Prevention Í5 (10): i829-34. Oct. 14,2003). Hoge, P 2005. CHP won't confiscate medicai marijuana; Lxjckyer reit- County of San Diego v. San Diego NORML. 2008. 165 Cai. App. 4th 798, erates that Prop. 215 permits usage. San Francisco Chronicie cert, denied, 556 U.S. __ (2009). Aug. 3O:B1. Doblin, R. & Kleiman, M. 1991. Marijuana as antiemetic medicine: A Institute of Medicine (iOM). i 999. Marijuana as Medicine: Assessing the survey of oncologists' experiences and attitudes. Journai of Ciinicai Science Base. Washington, DC: National Academy Press. Oncology 9(1): Í3Í4-Í9. irvine, G. 2006. Rural doctors' attitudes toward and knowledge of medic- Federal Bureau of investigation (FBi). 2009. Crime in America: FBI inal cannabis. Journal of Law and Medicine 14 (1): 135-42. Utiiform Crime Reports 2008. Washington, DC: U.S. Department of Kramer, RD. 1993. Listening to Prozac: A Psychiatrist Explores Justice. Available at http://www2.fbi.gov/ucr/cius2009/index.html. Antidepressant Drugs and the Re-making of the Self. New York: Gaoni, Y. & Mechoulam, R. 1964. isolation, structure, and partial syn- Viking. thesis of the active constituent of hashish. Joumal of the American L^ehmann, A.; Biackshaw L.A.; Branden, L.; Carisson, A.; Jensen J.; Chemical Society 86 (8): 1646-1647. Nygren, E. & Smid, S.D. 2002. Cannabinoid receptor agonism

Journal of Psychoactive Drugs 134 Volume 43 (2), April - June 201 i Reinarman et al. Who Are Medical Marijuana Patients?

inhibits transient lower esophageal sphincter relaxations and reñux Reinarman, C. & Levine, H.G. (Eds.) 1997. Crack In America: Demon in dogs. Ga.itroenterology 123 (6): 1129-34. Drugs and Social Justice. Berkeley: University of Califomia Martin, B.I.; Deyo, R.A.; Mirza, S.K.; Tumer, J.A.; Comstock, B.A.; Press. Hollingworth, W. & Sullivan, S.D. 2008. Expendituœs and health Samuels, D. 2008. Dr. Kush: How medical marijuana is transforming the status among adults with back and neck problems. Journal of the pot industry. New Yorker July 28: 49-62. American Medical Association 299 (6): 656-64. San Francisco Chronicle. 1996. State propositions. San Francisco McCarthy, K.T. 2004. Conversations about medical marijuana between Chronicle Nov. 7. A9. physicians and their patients. Journal of Legal Medicine 25 (3): Substance Abuse and Mental Health Services Administration 333-49. (SAMHSA), Office of Applied Studies. 2010. Results from the Mikos, R.A. 2009. On the limits of supremacy: Medical marijuana and the 2008 National Survey on Drug Use and Health: National Findings, states' overiooked power to legalize federal crime. Vanderbilt Law Table 1.24A. Rockville, MD: SAMHSA. Available at: http:// Review 62 (5): 1421-82. www.oas.samhsa.gov/nsduh/2k6nsduh/tabs/Sect 1 peTabs24to28.pdf Moore, S. 2009. Dispensers of marijuana find relief in policy shift. New Substance Abuse and Mental Health Services Administration, Office York Times Mm. 20: A\5. of Applied Statistics (SAMHSA). 2007. Results from the 2006 Murray, T.H.; Gaylin, W. & Macklin, R. 1984. Feeling Good and Doing National Survey on Drug Use and Health: National Findings. Belter: Ethics and Non-therapeutic Drug Use. Clifton, NJ: Humana Washington, DC: U.S. Dept. of Health and Human Services. Press. Substance Abuse and Mental Health Services Administration, Office of National Organization for the Reform of Marijuana Laws, Califomia Applied Statistics (SAMHSA). 2005. Table I.80B, Marijuana use Chapter (NORML). 2007. California Dispensary Locator. Available in lifetime, past year, and past month among persons aged 18-25 at http://www.canorml.org/prop/cbclist.htm). by racial/ethnic subgroup. In: National Survey on Drug Use and Nicoll, R. & Alger, B.E. 2004. The brain's own marijuana. Scientific Health: Detailed Tables. Washington, DC: U.S. Department of American December: 60-75. Health and Human Services. O'Connell, T.J. & Bou-Matar, C.B. 2007. Long term marijuana Tan, W.C; Lo, C; Jong, A.; Xing, L.; Fitzgerald, M.J.; Vollmer, W.M.; users seeking medical cannabis in California (2001-2007): Buist, S.A. & Sin, D.D. 2009. Marijuana and chronic obstruc- Demographics, social characteristics, patterns of cannabis and other tive lung disease: A population-based study. Canadian Medical drug use of 4117 applicants. Harm Reduction Journal 4: 16. Association Journal 180 (8): 814-20. Oregon Department of Human Services. 2008. Oregon Medical U.S. Œnsus Bureau. 2007. Population estimates. Available at Marijuana Program Statistics. Available at http://oregon.gov/ http://www.census.gov/popest/states/NST-ann-est2007.html DHS/ph/ommp/data.shtml. Ware, M.A.; Adams, H. & Guy, G.W. 2005. The medicinal use of cannabis Pletcher, M.J.; Kertesz, S.G.; Kohn, M.A. & Gonzales, R. 2008. Trends in the UK: Results of a nationwide survey. Journal of Clinical in opioid prescribing by race/ethnicity for patients seeking care Practice 59 0): 291-95. in U.S. emergency departments. Journal of the American Medical Woolrldge, E.; Barton, S.; Samuel, J.; Osorio, J.; Dougherty A. & Association 299 (1): 70-78. Holdcroft A. 2005. Cannabis use in HIV for pain and other medi- Reiman, A. 2007a. Medical cannabis patients: Patient profiles and health cal symptoms. Journal of Pain and Symptom Management 29 (4): care utilization pattems. Complementary Health Practice Review 12 358-67. (I): 31-50. Zeese, K.B. 1999. History of medical marijuana policy in US. Reiman, A. 2007b. Self-efficacy, social support and service integration International Journal of Drug Policy 10: 319-28. at medical cannabis facilities in the San Francisco Bay Area of Califomia. Health andSocial Care in the Community 16 (I): 31-41.

Journal of P.iychoactive Drugs 135 Volutne 43 (2), April - June 2011 Copyright of Journal of Psychoactive Drugs is the property of Haight Ashbury Publications and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. CASE REPORTS Effectiveness of Cannabidiol Oil for Pediatric Anxiety and Insomnia as Part of Posttraumatic Stress Disorder: A Case Report

Scott Shannon, MD, ABIHM; Janet Opila-Lehman, ND Perm J 2016 Fall;20(4):16-005 E-pub: 10/12/2016 http://dx.doi.org/10.7812/TPP/16-005

ABSTRACT of medical marijuana, which do contain activating adenosine receptors which play Introduction: Anxiety and sleep THC, such as “Charlotte’s Web.” a significant role in cardiovascular function disorders are often the result of posttrau- The most abundant compound in can- and cause a broad anti-inflammatory effect matic stress disorder and can contribute nabis, THC is also a cannabinoid. The throughout the body.5 At high concentra- to an impaired ability to focus and to THC component induces the psychoac- tions, CBD directly activates the 5-HT1A demonstration of oppositional behaviors. tive effect, “high.” A cannabis plant has serotonin receptor, thereby conferring an Case Presentation: These symptoms different amounts of CBD and THC antidepressant effect.6 Cannabidiol has were present in our patient, a ten-year- depending on the strain and thus provides been found to be an antagonist at the po- old girl who was sexually abused and different recreational or medicinal effects. tentially new third cannabinoid receptor, had minimal parental supervision as a The cannabinoid profile of industrial hemp GPR55, in the caudate nucleus and puta- young child under the age of five. Phar- or medical marijuana is ideal for people men, which if stimulated may contribute maceutical medications provided partial looking for the medical benefits of CBD to osteoporosis.7 relief, but results were not long-lasting, without the “high” of the THC. Since the 1940s, a considerable number and there were major side effects. A The mechanism of action of CBD is of published articles have dealt with the trial of cannabidiol oil resulted in a multifold.1-3 Two cannabinoid receptors chemistry, biochemistry, pharmacology, maintained decrease in anxiety and a are known to exist in the human body: and clinical effects of CBD.8 The last de- steady improvement in the quality and CB1 and CB2 receptors. The CB1 recep- cade has shown a notable increase in the quantity of the patient’s sleep. tors are located mainly in the brain and scientific literature on CBD, owing to its Discussion: Cannabidiol oil, an in- modulate neurotransmitter release in a identification for reducing nausea and creasingly popular treatment of anxiety manner that prevents excessive neuronal vomiting, combating psychotic disorders, and sleep issues, has been documented activity (thus calming and decreasing reducing inflammation, decreasing anxi- as being an effective alternative to phar- anxiety), as well as reduces pain, reduces ety and depression, improving sleep, and maceutical medications. This case study inflammation, regulates movement and increasing a sense of well-being.9-12 Find- provides clinical data that support the posture control, and regulates sensory ings presented at the 2015 International use of cannabidiol oil as a safe treatment perception, memory, and cognitive func- Cannabinoid Research Society at its 25th for reducing anxiety and improving tion.a2 An endogenous ligand, anan- Annual Symposium reported the use of sleep in a young girl with posttraumatic damide, which occurs naturally in our CBD as beneficial for kidney fibrosis and stress disorder. bodies, binds to the CB1 receptors through inflammation, metabolic syndrome, over- the G-protein coupling system. CBD has weight and obesity, anorexia-cachexia syn- INTRODUCTION an indirect effect on the CB1 receptors drome, and modification of osteoarthritic Cannabidiol (CBD) oil is a naturally oc- by stopping the enzymatic breakdown of and other musculoskeletal conditions.13-16 curring constituent of industrial hemp and anandamide, allowing it to stay in the sys- Although studies have demonstrated the marijuana, which are collectively called tem longer and provide medical benefits.4 calming, anti-inflammatory, and relaxing cannabis. CBD oil is 1 of at least 85 can- CBD has a mild effect on the CB2 recep- effects of CBD, clinical data from actual nabinoid compounds found in cannabis tors, which are located in the periphery in cases is minimal. This case study offers and is popular for its medicinal benefits. lymphoid tissue. CBD helps to mediate the evidence that CBD is effective as a safe After tetrahydrocannabinol (THC), CBD release of cytokines from the immune cells alternative treatment to traditional psy- oil is the second-most-abundant compo- in a manner that helps to reduce inflam- chiatric medications for reducing anxiety nent of cannabis. Other names for CBD oil mation and pain.2 and insomnia.17 include CBD-rich hemp oil, hemp-derived Other mechanisms of action of CBD CBD oil, or CBD-rich cannabis oil. Con- include stimulation of vanilloid pain recep- CASE PRESENTATION sidered to be generally safe, CBD has been tors (TRPV-1 receptor), which are known A ten-year-old girl presented in Janu- used medicinally for decades. However, to mediate pain perception, inflamma- ary 2015 for a reevaluation of behaviors CBD is not medical marijuana and should tion, and body temperature.5 In addition, related to her diagnosis of posttraumatic be distinguished from high-CBD strains CBD may exert its anti-anxiety effect by stress disorder (PTSD) secondary to sexual

Scott Shannon, MD, ABIHM, is an Assistant Clinical Professor of Psychiatry at the University of Colorado School of Medicine in Fort Collins. E-mail: [email protected]. Janet Opila-Lehman, ND, is a Naturopathic Physician at the Wholeness Center in Fort Collins, CO. E-mail: [email protected].

108 The Permanente Journal/Perm J 2016 Fall;20(4):16-005 CASE REPORTS Effectiveness of Cannabidiol Oil for Pediatric Anxiety and Insomnia as Part of Posttraumatic Stress Disorder: A Case Report

abuse. Her chief issues included anxiety, Her father had died 6 months earlier in a marijuana her entire pregnancy with the insomnia, outbursts at school, suicidal motor vehicle accident, and our patient’s girl. The patient presented in January 2012 ideation, and self-destructive behaviors. maternal grandparents became her perma- as displaying aggressive, disobedient, im- Her grandmother, who has permanent nent guardians. Before her father’s death, pulsive, and sexually inappropriate behav- custody of the patient and her younger our patient had no supervision from her iors. She also demonstrated low self-esteem brother, accompanied her. father and very little supervision from her and anxiety and had poor sleep (restless, Our patient had been seen for an initial mother. An 11-year-old boy had molested interrupted, and unable to sleep alone). evaluation in January 2012 and received her when she was 3 years old. Her medi- Workup during 2012 included labora- a diagnosis of PTSD secondary to sexual cal history included her mother having tory studies, which ruled out a thyroid abuse on the basis of her history, clinical methadone addiction, alcoholism, bipolar dysfunction and an iron or vitamin D observations, and behaviors (Table 1). disorder, and depression. Her mother used deficiency. The patient was started on a

Table 1. Timeline Date Presentation Medications Supplements Other January 31, New evaluation: 7.5-year-old girl. History None Melatonin, 1 mg/night February 14, 2012, laboratory values: 2012 of sexual abuse and neglect. Issues: TSH, 2.46 mIU/L (reference range, 0.47- Insomnia, sexual behaviors. Diagnosis: 4.68 mIU/L); ferritin: 21 ng/mL (reference PTSD secondary to sexual abuse. range, 10-150 ng/mL). February 16, 2012, laboratory values:

Vitamin D3: 39 ng/mL (reference range, 20-50 ng/mL) February 20, Sleeping 2-3 hours/night. Started counseling; , 0.05 Inositol, 3 g 3 times/d; EPA fish Eye movement desensitization and 2012 Cooperative and good behavior at counseling mg (half tablet) at oil, 500 mg/d reprocessing therapy recommended session. Anxious, traumatized. bedtime February 22, Did not do well with clonidine because of Started March 7, 2012: ECG was normal 2012 hallucinations, so she discontinued that imipramine treatment. Behavior still very rough; sleep therapy, 25 mg poor. at bedtime August 8, Good summer. In play therapy. Overall Imipramine, 2012a better sleep and energy with imipramine 25 mg at bedtime therapy. Patient’s 6-year-old brother also now in therapy. January 21, Returned for evaluation and treatment after Off all Melatonin, 5 mg; St John’s 2015 3 years. Suicidal ideation; cut self on leg; medications for wort, 450 mg twice/d; defiant and stubborn. Had psychotherapy past 18 months magnesium, 300 mg/d; 3 years straight twice a month. Sleeps with diphenhydramine, 25 mg/night brother; can’t sleep alone. February 16, Hard to manage. Has outbursts at school. Magnesium and St John’s February 11, 2015: Normal cortisol and 2015 wort: stopped treatment; EPA DHEA levels fish oil, 750 mg/d; diphenhydramine, 25 mg/night March 16, Better overall. Started animal-assisted EPA fish oil, 750 mg/d; Started a regimen of CBD oil, 25 mg 2015 therapy. diphenhydramine, 25 mg/night (1 capsule)/d at 6 pm April 14, 2015 Sleeping better with CBD treatment. Getting EPA fish oil, 750 mg/d; CBD oil, 25 mg (1 capsule)/d at 6 pm biofeedback. Has stomachaches. Mood is diphenhydramine, 25 mg/night more at ease. May 26, 2015 “Ghosts” waking patient up at night. EPA fish oil, 750 mg/d CBD oil, 25 mg (1 capsule)/d at 6 pm July 22, 2015 Sleeping better; able to sleep in own room EPA fish oil, 750 mg/d CBD liquid, 12 mg (in 4 sublingual sprays)/ 3-4 nights/wk. night; 12 mg more (in 4 sublingual sprays) during the day as needed for anxiety, typically 3 or 4 times/wk August 24, Sleeping well. Handling school well. EPA fish oil, 750 mg/d CBD oil, 25 mg (1 capsule)/night; CBD 2015 liquid, 6-12 mg (in 2-4 sublingual sprays) as needed for anxiety, typically 2 or 3 times/wk a There were additional visits in 2012 with no substantial changes. CBD = cannabidiol; DHEA = dehydroepiandrosterone; ECG = electrocardiogram; EPA = eicosapentaenoic acid; PTSD = posttraumatic stress disorder; TSH = thyroid stimulating hormone.

The Permanente Journal/Perm J 2016 Fall;20(4):16-005 109 CASE REPORTS Effectiveness of Cannabidiol Oil for Pediatric Anxiety and Insomnia as Part of Posttraumatic Stress Disorder: A Case Report

regimen of 1 mg/night of melatonin, which Anxiety Related Disorders demonstrated was ultimately able to sleep through the helped her sleep duration. Three grams of an improvement (Table 2). night most nights in her own room, was inositol 3 times a day and 500 mg/d of A trial of CBD supplements (25 mg) less anxious at school and home, and dis- eicosapentaenoic fish oil were also helpful was then initiated at bedtime, and 6 mg played appropriate behaviors. The patient’s in reducing her anxiety. A trial of cloni- to 12 mg of CBD sublingual spray was grandmother (her caregiver) reported: “My dine was implemented, which resulted in administered during the day as needed for granddaughter’s behaviors are definitely hallucinations and thus was discontinued. anxiety. A gradual increase in sleep quality better being on the CBD. Her anxiety is The patient was switched to a regimen of and quantity and a decrease in her anxiety not gone, but it is not as intense and she is 25 mg of imipramine at bedtime to de- were noted. After 5 months, the patient much easier to be around. She now sleeps crease her anxiety, which appeared to be was sleeping in her own room most nights in her own room most of the time, which helpful. Counseling sessions were started. and handling the new school year with no has never happened before.” The patient continued psychotherapy for difficulties. No side effects were observed Further study will need to be conducted 3 years, but she was not seen again in our from taking the CBD oil. to determine the permanency of our pa- clinic until the return visit in January tient’s positive behaviors and how long she 2015, when she was not receiving any of DISCUSSION will need to continue taking the CBD oil. her medications and supplements. Studies repeatedly recognize the preva- We do not have a reasonable foundation lence of an anxiety-provoked sleep disor- to recommend dosing from the scientific 20 CBD oil can be an effective der after a traumatic experience. Our literature. However, in our experience, patient was definitely experiencing this this supplement given 12 mg to 25 mg compound to reduce anxiety and phenomenon, which was aggravated by once daily appears to provide relief of insomnia secondary to PTSD. daily stressful activities. key symptoms with minimal side effects. The main finding from this case study Our patient did not voice any complaints At the patient’s return in January 2015, is that CBD oil can be an effective com- or discomfort from the use of CBD. We she demonstrated the same prominent pound to reduce anxiety and insomnia routinely asked about headache, fatigue, symptoms as at her initial presentation. secondary to PTSD. A review of the lit- and change in appetite or agitation in ad- At that time, the initial treatment in- erature suggests some benefits from the dition to conducting a routine psychiatric cluded the following supplements and use of CBD because of its anxiolytic and evaluation. Although CBD is considered medications to assist with her sleep and sleep-inducing effects.9 Animal studies generally safe,17 the long-term effects are anxiety: melatonin, 5 mg/night; magne- support use of this treatment and report yet to be studied. sium, 300 mg/d; and diphenhydramine that “CBD may block anxiety-induced The ultimate goal is to gradually taper (Benadryl), 25 mg/night. Our patient [rapid eye movement] sleep alteration via her off the use of CBD oil and transition demonstrated slight gains but was still hav- its anxiolytic effect on the brain.”21 our patient into lifelong coping strategies ing outbursts at school and was reportedly The strength of this particular case is such as yoga, meditation, and various other difficult to manage at home. In addition, that our patient was receiving no phar- therapeutic activities. v her underlying anxiety continued. maceutical medications (other than non- Cannabidiol oil was explored as a po- prescription diphenhydramine) but only a GW Pharmaceuticals is the founder of the Cannabinoid Research Institute, directed by Philip Robson, MD. tential additional treatment to help her nutritional supplements and the CBD Further research articles listed. insomnia and anxiety, but we deferred oil to control her symptoms. Her scores for two months while we waited for a on the sleep scale and the anxiety scale Disclosure Statement response from other interventions. The consistently and steadily decreased during The author(s) have no conflicts of interest to grandmother preferred reducing the phar- a period of 5 months (see Table 2). She disclose. macologic load given her granddaughter’s failure to respond long term to psychiatric Table 2. Patient’s clinical progress in Acknowledgments medications. sleep and anxiety CannaVest Corp, San Diego, CA, which had In March 2015, CBD oil was recom- no involvement in the case study or distribution Sleep scale SCARED of the product, provided the CBD oil that was mended as a potential additional treatment a b Date of visit score score administered to the patient. No financial support to help her insomnia and anxiety, and March 16, 2015 59 34 was provided. her grandmother provided full informed May 25, 2015 42 24 Kathleen Louden, ELS, of Louden Health consent. Our patient was administered the Communications provided editorial assistance. July 22, 2015 41 19 Sleep Disturbance Scale for Children18 and August 24, 2015 37 16 the Screen for Anxiety Related Disorders How to Cite this Article September 22, 2015 38 18 (SCARED)19 before taking the CBD oil Shannon S, Oplia-Lehman J. Effectiveness of a and each month afterward for the next 5 A score of more than 50 is considered indicative of cannabidiol oil for pediatric anxiety and insomnia as a sleep disorder on the Sleep Disturbance Scale for part of posttraumatic stress disorder: A case report. months. Test scores on the Sleep Distur- Children. Perm J 2016 Fall;20(4):16-005. DOI: http://dx.doi. b bance Scale for Children and Screen for A SCARED score over 25 indicates a high probability org/10.7812/TPP/16-005. of a childhood anxiety disorder. SCARED = Screen for Anxiety Related Disorders.

110 The Permanente Journal/Perm J 2016 Fall;20(4):16-005 CASE REPORTS Effectiveness of Cannabidiol Oil for Pediatric Anxiety and Insomnia as Part of Posttraumatic Stress Disorder: A Case Report

References 8. Zhornitsky S, Potvin S. Cannabidiol in humans— 16. Liu A. Therapeutic efficacy of a peripherally restricted 1. Campos AC, Moreira FA, Gomes FV, Del Bel EA, the quest for therapeutic targets. Pharmaceuticals CB1R antagonist/AMPK activator in diet-induced Guimarães FS. Multiple mechanisms involved in the (Basel) 2012 May 21;5(5):529-52. DOI: http://dx.doi. obesity/metabolic syndrome. Proceedings of the large-spectrum therapeutic potential of cannabidiol org/10.3390/ph5050529. 25th Anniversary Symposium of the International in psychiatric disorders. Philos Trans R Soc Lond B 9. Zuardi AW. Cannabidiol: from an inactive cannabinoid Cannabinoid Research Society; 2015 Jun 28-Jul 3; Biol Sci 2012 Dec 5;367(1607):3364-78. DOI: http:// to a drug with wide spectrum of action. Rev Bras Wolfville, Nova Scotia, Canada. dx.doi.org/10.1098/rstb.2011.0389. Psiquiatr 2008 Sep;30(3):271-80. DOI: http://dx.doi. 17. Bergamaschi MM, Queiroz RH, Zuardi AW, 2. Mechanism of action [Internet]. Cambridge, United org/10.1590/s1516-44462008000300015. Crippa JA. Safety and side effects of cannabidiol, Kingdom: GW Pharmaceuticals plc; c2014 [cited 10. Burstein S. Cannabidiol (CBD) and its analogs: a a Cannabis sativa constituent. Curr Drug Saf 2015 Aug]. Available from: www.gwpharm.com/ review of their effects on inflammation. Bioorg Med 2011 Sep 1;6(4):237-49. DOI: http://dx.doi. mechanism-of-action.aspx. Chem 2015 Apr 1;23(7):1377-85. DOI: http://dx.doi. org/10.2174/157488611798280924. 3. McPartland JM, Guy G. The evolution of cannabis org/10.1016/j.bmc.2015.01.059. 18. Ferreira VR, Carvalho LB, Ruotolo F, de Morais JF, and coevolution with the cannabinoid receptor—a 11. Fernández-Ruiz J, Sagredo O, Pazos MR, et al. Prado LB, Prado GF. Sleep disturbance scale hypothesis. In: Guy GW, Whittle BA, Robson Cannabidiol for neurodegenerative disorders: for children: translation, cultural adaptation, and PJ, editors. The medicinal uses of cannabis and important new clinical applications for this validation. Sleep Med 2009 Apr;10(4):457-63. DOI: cannabinoids. 1st ed. London, United Kingdom: phytocannabinoid? Br J Clin Pharmacol 2013 http://dx.doi.org/10.1016/j.sleep.2008.03.018. Pharmaceutical Press; 2004. p 71-102. Feb;75(2):323-33. DOI: http://dx.doi.org/10.1111/ 19. Birmaher B, Khetarpal S, Cully M, Brent D, 4. Leweke FM, Piomelli D, Pahlisch F, et al. Cannabidiol j.1365-2125.2012.04341.x. McKenzie S. Screen for child anxiety related enhances anandamide signaling and alleviates 12. Zuardi AW, Crippa JA, Hallak JE, Moreira FA, disorders (SCARED) [Internet]. Pittsburgh, PA: psychotic symptoms of schizophrenia. Transl Guimarães FS. Cannabidiol, a Cannabis sativa Western Psychiatric Institute and Clinic, University Psychiatry 2012 Mar 20;2:e94. DOI: http://dx.doi. constituent, as an drug. Braz J Med of Pittsburgh; 1995 Oct [cited 2016 Apr 26]. org/10.1038/tp.2012.15. Biol Res 2006 Apr;39(4):421-9. DOI: http://dx.doi. Available from: www.pediatricbipolar.pitt.edu/content. 5. Lee MA. CBD: how it works. O’Shaughnessy’s org/10.1590/s0100-879x2006000400001. asp?id=2333#3304. [Internet] 2011 Autumn [cited 2016 Apr 26]:14. 13. Fingerle J. CB2 agonism protects from inflammation 20. Pace-Schott EF, Germain A, Milad MR. Sleep and Available from: www.os-extra.cannabisclinicians.org/ related kidney damage and fibrosis. Proceedings of REM sleep disturbance in the pathophysiology of wp-content/uploads/2012/07/CBDiary21.pdf. the 25th Anniversary Symposium of the International PTSD: the role of extinction memory. Biol Mood 6. Crippa JA, Derenusson GN, Ferrari TB, et al. Neural Cannabinoid Research Society; 2015 Jun 28-Jul 3; Anxiety Disord 2015 May 29;5:3. DOI: http://dx.doi. basis of anxiolytic effects of cannabidiol (CBD) in Wolfville, Nova Scotia, Canada. org/10.1186/s13587-015-0018-9. generalized social anxiety disorder: a preliminary 14. Purohit V. Role of cannabinoids in chronic pain. 21. Hsiao YT, Yi PL, Li CL, Chang FC. Effect of report. J Psychopharmacol 2011 Jan;25(1):121-30. Proceedings of the 25th Anniversary Symposium cannabidiol on sleep disruption induced by the DOI: http://dx.doi.org/10.1177/0269881110379283. of the International Cannabinoid Research Society; repeated combination tests consisting of open field 7. McHugh D, Tanner C, Mechoulam R, Pertwee RG, 2015 Jun 28-Jul 3; Wolfville, Nova Scotia, Canada. and elevated plus-maze in rats. Neuropharmacology Ross RA. Inhibition of human neutrophil 15. Starowicz K. Role of endocannabinoid system in 2012 Jan;62(1):373-84. DOI: http://dx.doi. chemotaxis by endogenous cannabinoids and pathogenesis of osteoarthritic pain. Proceedings of org/10.1016/j.neuropharm.2011.08.013. phytocannabinoids: evidence for a site distinct from the 25th Anniversary Symposium of the International CB1 and CB2. Mol Pharmacol 2008 Feb;73(2): Cannabinoid Research Society; 2015 Jun 28-Jul 3; 441-50. DOI: http://dx.doi.org/10.1124/ Wolfville, Nova Scotia, Canada. mol.107.041863.

Marijuana and Medicine

Scientific data indicate the potential therapeutic value of cannabinoid drugs, primarily [tetrahydrocannabinol], for pain relief, control of nausea and vomiting, and appetite stimulation; smoked marijuana, however, is a crude [tetrahydrocannabinol] delivery system that also delivers harmful substances.

— Joy JE, Watson SJ Jr, Benson JA Jr. Marijuana and medicine: Assessing the science base. Washington, DC: National Academies Press; 1999.

The Permanente Journal/Perm J 2016 Fall;20(4):16-005 111

Question 5

Letters of support provided by physicians with knowledge of the disease or condition. DocuSign Envelope ID: D6D51530-0821-4177-A1BA-100C86A3772A

To Whom It May Concern,

We, the undersigned physicians, support adding insomnia to the qualifying conditions list under Ohio’s Medical Marijuana Control Program.

After a thorough review of the available science, research and information on treating insomnia with medical marijuana, we believe it to be an effective treatment, and that the benefits outweigh the risks.

Sleep is vital to a person’s health. Diseases such as insomnia, which prevent a patient from getting a full nights rest, can lead to serious issues including heart disease, high blood pressure, and mental illness. Successful treatment of insomnia when it is diagnosed is vital to the ongoing health of patients suffering from it.

The causes of insomnia are varied, and successful treatment of the disease must take that into account. It’s important that physicians be able to utilize a range of treatment options based on a patient’s unique circumstances.

We believe the evidence supports medical marijuana as a treatment option for insomnia, and we ask that the State Medical Board of Ohio add insomnia as a qualifying condition under Ohio’s Medical Marijuana Control Program.

Sincerely,

Daniel Neides

MD DocuSign Envelope ID: 7BEB1890-1114-412F-A006-3B1266040D9F

To Whom It May Concern,

We, the undersigned physicians, support adding insomnia to the qualifying conditions list under Ohio’s Medical Marijuana Control Program.

After a thorough review of the available science, research and information on treating insomnia with medical marijuana, we believe it to be an effective treatment, and that the benefits outweigh the risks.

Sleep is vital to a person’s health. Diseases such as insomnia, which prevent a patient from getting a full nights rest, can lead to serious issues including heart disease, high blood pressure, and mental illness. Successful treatment of insomnia when it is diagnosed is vital to the ongoing health of patients suffering from it.

The causes of insomnia are varied, and successful treatment of the disease must take that into account. It’s important that physicians be able to utilize a range of treatment options based on a patient’s unique circumstances.

We believe the evidence supports medical marijuana as a treatment option for insomnia, and we ask that the State Medical Board of Ohio add insomnia as a qualifying condition under Ohio’s Medical Marijuana Control Program.

Sincerely,

Cynthia Taylor

do