Psychopharmacology (2005) 178: 259–267 DOI 10.1007/s00213-004-2009-1
ORIGINA L IN VESTI GATION
Miriam Z. Mintzer . Roland R. Griffiths
Flumazenil-precipitated withdrawal in healthy volunteers following repeated diazepam exposure
Received: 31 March 2004 / Accepted: 9 August 2004 / Published online: 25 September 2004 # Springer-Verlag 2004
Abstract Rationale: Parametric preclinical studies of the benzodiazepine antagonist flumazenil have contributed to the understanding of the physical dependence associated with chronic benzodiazepine use. However, few parametric studies have been conducted in human participants. Objective: This study was designed to assess the effect of duration of benzodiazepine exposure on the intensity of flumazenil-precipitated withdrawal in healthy volunteers. Method: Participants were randomly assigned to receive either oral diazepam (15 mg/70 kg; n=10) or placebo (n=8) capsules nightly for 28 days. Effects of flumazenil (1 mg/70 kg, intravenously administered) were assessed in challenge sessions conducted before capsule ingestion, and after 1, 7, 14, and 28 days of capsule ingestion. Results: Flumazenil produced a profile of participantrated effects consistent with benzodiazepine withdrawal that peaked immediately after completion of the 5-min flumazenil injection and rapidly dissipated thereafter. The magnitude of these effects was comparable after 7, 14, and 28 days of diazepam. Flumazenil also produced modest elevations in blood pressure and decreases in skin temperature in the diazepam group, both of which were sustained throughout the approximate 60-min session. Conclusions: These findings support previous human research studies indicating that flumazenil precipitates withdrawal after short chronic exposure to benzodiazepines and suggests that duration of exposure does not influence the intensity of withdrawal beyond the first week of exposure.
M. Z. Mintzer (*) . R. R. Griffiths Johns Hopkins University School of Medicine, 5510 Nathan Shock Drive, Baltimore, MD 21224, USA e-mail:
[email protected] Tel.: +1-410-5500529 Fax: +1-410-5500030
Introduction Benzodiazepines, typically prescribed to treat anxiety and sleep disorders, are among the most commonly prescribed psychotropic medications (Bruce et al. 2003; Kaufman et al. 2002). In 2003 in the United States, benzodiazepines were prescribed at a rate of 72 million prescriptions per year, with the only psychotropics exceeding this rate being codeine (and combination products) and the serotonin reuptake inhibitors (SSRS/SNRI) (IMS Health 2003). The efficacy of long-term treatment with benzodiazepines has not been demonstrated, and the risk of physical dependence has long been recognized as a significant concern associated with benzodiazepine use (Griffiths and Weerts 1997). Guidelines for physicians typically suggest prescribing a short-term course of benzodiazepines (Physicians’ Desk Reference 2003). Despite this, benzodiazepines are often used on a long-term basis: in a 1990 survey of the general population in the United States, 25% of past-year users of anxiolytics and 14% of past-year users of hypnotics reported using these drugs continuously for 1 year or more (Griffiths and Weerts 1997; Woods et al. 1992). Chronic use of benzodiazepines at therapeutic doses can produce physical dependence, as reflected in the appearance of a unique constellation of symptoms when the drug is abruptly withdrawn. Symptoms of benzodiazepine withdrawal include anxiety, dizziness, nausea, lack of coordination, heightened sensory perception, depersonalization, blurred vision, irritability, insomnia, palpitations, muscle tension, tremors, dysphoria, and headache; in more severe cases, they may also include seizures, paresthesias, perceptual distortions, delirium, panic attacks, and psychotic episodes. Flumazenil (Romazicon), a benzodiazepine receptor antagonist, has been used extensively in preclinical research to precipitate benzodiazepine withdrawal. In general, the symptoms of withdrawal precipitated by flumazenil are similar to the symptoms observed when a benzodiazepine agonist is withdrawn naturally (Woods et al. 1987) and, for this reason, flumazenil has been a useful tool in studies evaluating the conditions
260
under which benzodiazepine physical dependence is likely to develop. Preclinical studies of flumazenil-precipitated withdrawal have demonstrated that the dose of the benzodiazepine agonist (Lukas and Griffiths 1984; Sloan et al. 1993), the specific pharmacokinetic profile of the agonist (Martin et al. 1990; Sloan et al. 1991b), the duration of exposure to the agonist (Lukas and Griffiths 1982, 1984; Rosenberg and Chiu 1985; Sloan et al. 1991a; Wilson and Gallager 1988; Zharkovskii and Zharkovskaia 1987 as cited in Woods et al. 1992; but cf. Kaminski et al. 2003), and the dose of flumazenil (Lamb and Griffiths 1987; Martin et al. 1990; McNicholas et al. 1988; Sannerud et al. 1991) all contribute to the intensity of withdrawal. Surprisingly few systematic studies of flumazenilprecipitated withdrawal have been conducted with human participants, despite the fact that such studies could potentially provide insight into the mechanisms underlying spontaneous benzodiazepine withdrawal and provide important clinical data on the time course of the development of physical dependence. Studies of flumazenil-precipitated withdrawal have been conducted in volunteers who have a history of therapeutic benzodiazepine use (Bernik et al. 1991, 1998; Harrison-Read et al. 1996; Mintzer et al. 1999). These studies have shown that flumazenil-precipitated withdrawal occurs after chronic use and, interestingly, even after chronic use of relatively low doses (Mintzer et al. 1999). Given that benzodiazepines are so commonly prescribed, understanding how the duration of agonist exposure affects the developmental course of physical dependence may be particularly important. To address this issue, studies in which healthy volunteers with no history of prior benzodiazepine use are exposed experimentally to a benzodiazepine are critical. To our knowledge, only two prior studies of flumazenilprecipitated withdrawal in healthy volunteers without histories of benzodiazepine use have explicitly addressed the issue of duration of agonist exposure. In one study, flumazenil (3 mg) did not precipitate withdrawal symptoms in 36 healthy volunteers who were pretreated with lormetazepam (2 mg) for 30 days (Gerra et al. 1996). This study also included a group of long-term dependent users who, surprisingly, also did not exhibit withdrawal after flumazenil administration. Although the reasons for the lack of withdrawal effects in this study are unclear, these results suggest that a 30-day exposure to lormetazepam may not be sufficient to induce physical dependence. A different outcome occurred in a prior study from our laboratory in which a group design was used to examine the influence of duration of lorazepam exposure on flumazenil-precipitated withdrawal in healthy volunteers without histories of benzodiazepine use (Griffiths et al. 1993). The findings indicated that flumazenil (3 mg) did not precipitate withdrawal in a group receiving a single exposure of lorazepam (3 mg), but that withdrawal symptoms were precipitated in groups exposed to 3, 7, and 14 days of lorazepam pretreatment. An unexpected finding was that the intensity of some of the precipitated withdrawal symptoms was lower in those who received 14
days of lorazepam exposure than in those who received 3 days or 7 days of exposure. These findings are in contrast to preclinical data which have generally shown that the severity of withdrawal symptoms is an increasing function of the duration of exposure to the agonist (Lukas and Griffiths 1982, 1984; Rosenberg and Chiu 1985; Sloan et al. 1991a; Wilson and Gallager 1988; Zharkovskii and Zharkovskaia 1987 as cited in Woods et al. 1992; but cf. Kaminski et al. 2003). However, due to the group design used in our prior study, it was not clear whether the unexpected result was due to unknown group-specific factors that were not controlled. The purpose of the present study was to assess the influence of duration of benzodiazepine exposure on flumazenil-precipitated withdrawal effects in healthy volunteers when exposure duration is varied within subjects. Participants received diazepam for 28 consecutive days, with flumazenil challenge sessions occurring before diazepam exposure and after 1, 7, 14, and 28 days of diazepam exposure. During these sessions, subjective, observer-rated, and physiological effects were assessed. Diazepam, which is slowly eliminated, was administered nightly in a single high therapeutic dose (15 mg/70 kg before bed). In addition, a control group received flumazenil challenge sessions on the same schedule, but after ingesting placebo capsules each night.
Methods Subjects Participants were 27 healthy adult volunteers (11 female) without histories of drug abuse. Nine participants (4 in the diazepam group; 5 in the placebo group) did not complete the study for reasons including medical issues (fainting when the catheter was placed, nausea) and noncompliance with study procedures. Participants were randomly assigned to the two groups with the constraint that the groups (diazepam group, n=10; placebo group, n=8) were balanced for age and baseline anxiety [using the StateTrait Anxiety Inventory (STAI); Spielberger et al. 1970]. The mean ages of the diazepam and placebo groups were 33.8 years and 30.4 years, respectively; the mean baseline STAI values were 30.4 and 29.3, respectively. With the exception of two participants in the diazepam group who reported no or very infrequent alcohol use, all participants reported regular social alcohol use (at least monthly). Three participants in the diazepam group and two in the placebo group reported smoking cigarettes. No participants reported history of benzodiazepine use. Participants were required to abstain from the use of alcohol throughout the study and from the use of caffeine and food for 2 h before each experimental session. Before each session, participants were tested for the presence of drugs in urine (benzodiazepines, barbiturates, opioids, amphetamines, and cocaine) using an EMIT system (Syva Co., Palo Alto, CA, USA) and the presence of alcohol in expired air using a breathalyzer test. In addition, urine
261
pregnancy tests were conducted before each session for female participants, none of which was positive. Participants were informed that, during their participation in the study, they would receive various drugs orally and intravenously (IV), and that these drugs could include placebo, various sedatives, anxiolytics, stimulants and weight-loss medications. In addition, participants were specifically informed that one of the drugs they would receive was flumazenil and that the drug sometimes causes unpleasant reactions in people taking benzodiazepines. This study was approved by the Institutional Review Board of the Johns Hopkins Bayview Medical Center. Participants gave their written informed consent before beginning the study and were paid for their participation.
capsule filling. Size O, opaque, positive locking gelatin capsules (Elanco Qualicaps Indianapolis, Indiana) served as the containers for both diazepam and placebo doses. Finished doses were packaged in sealed unit dose packages with appropriate labeling (Regional Service Center, Woburn, MA). Each flumazenil IV dose (1.0 mg/70 kg) was prepared on a mg/70 kg/4 ml basis and filtered through a sterile pyrogen free 0.22 μm-MillexGS Millipore filter unit (Millipore Products Division, Bedford, MA, USA). Doses were prepared by diluting flumazenil (commercially available 1 mg/10 ml solution; Romazicon; Roche Laboratories, a division of Hoffman La Roche, Nutley, NJ, USA) with 0.9% sodium chloride. Drug administration
General procedures All participants completed five experimental sessions as outpatients at the Behavioral Pharmacology Research Unit. Sessions involved IV flumazenil administration (1.0 mg/70 kg) and completion of experimental measures. The first experimental session occurred on the day before participants began ingesting nightly capsules (day 0), and the next four sessions occurred after 1, 7, 14, and 28 days of nightly capsule ingestion. Participants in the diazepam group received a nightly capsule containing 15 mg/70 kg diazepam, and those in the placebo group received a nightly capsule containing lactose. In order to determine plasma levels of diazepam and its metabolite nordiazepam, 5 ml plasma was obtained from each participant at the beginning of each experimental session. The tube was then centrifuged at 4°C at 1,809 g for 10 min. The plasma fraction was pipetted off and dispensed to a plastic vial. The samples were kept frozen at −70°C until analyses were conducted (Quest Diagnostics, Baltimore, MD). Analyses were only conducted on the samples from the diazepam group. For 2 weeks after completing the final experimental session, participants were required to maintain regular telephone contact with one of the investigators. Participants were interviewed regarding symptoms associated with benzodiazepine withdrawal using a previously developed clinical scale assessing benzodiazepine withdrawal (Busto et al. 1989). This scale consists of 20 items that participants rated on a scale from 0 to 4 based on intensity and two sleep-related items: total hours of sleep the preceding night and latency (in minutes) to fall asleep the preceding night. Drug preparation Diazepam take-home capsules (diazepam group; 15 mg/70 kg) were compounded from crushed diazepam 10-mg tablets (Roche Laboratories, Inc., Nutley, NJ 07110). Lactose monohydrate, NF powder (Amend Drug and Chemical Company, Irvington, NJ) was used as the placebo (placebo group) and the diluent for diazepam
After the first session (day 0), participants were given the take-home capsules on a weekly basis. On day 21 of the study, on which no experimental session was scheduled, participants came to the laboratory to pick up the next week’s doses. Participants were instructed to take one take-home capsule each night immediately before going to bed and were given standard instructions and precautions for nightly hypnotic use (Physician’s Desk Reference 2003). Although participants were not observed taking their nightly capsules, compliance with the protocol was maximized by having participants use an automated telephone voice mail system to provide a date and time stamp of the time each nightly capsule was taken, and by having participants provide urine and blood samples as described above. During the experimental sessions, flumazenil was administered by a physician via an IV catheter inserted into the right arm prior to each session. A slow-drip IV line was maintained during each session. In order to ensure safety, dose administration occurred over a period of 5 min (Physician’s Desk Reference 2003); accordingly, an initial dose of 0.2 mg/70 kg (4 ml total volume) was administered over 15 s and additional doses of 0.2 mg/70 kg each (4 ml total volume each) were administered at 60-s intervals up to a maximum dose of 1.0 mg/70 kg (20 ml total volume) in 5 min. During the flumazenil injection, participants were asked repeatedly how they were feeling. They were told that they could ask for the injection to stop at any point during administration if they felt uncomfortable. The protocol stipulated that the injection be terminated if a significant adverse event occurred or at the request of the participant. No significant adverse events occurred and the injection was not terminated prematurely for any participant (i.e., all participants received the full 1.0 mg/70 kg dose). The physician remained in the room for 15 min after each flumazenil injection was completed and monitored participants periodically throughout each session.
262
Experimental measures All participant and observer ratings were completed 20 min and 10 min before (baseline) the 5-min flumazenil injection, immediately after completion of the injection, and 20, 40, and 60 min after completion of the injection. In addition, immediately upon completion of the 60-min ratings, participants completed all participant ratings again with instructions to respond to the questions retrospectively based on the maximal effect experienced during the session. Physiological measures were taken continuously from 20 min before the injection until the end of the experimental session. Participant ratings Participant ratings were completed on the computer and consisted of four visual analog scales (cf. “Results”), a computerized version of the standardized STAI state version (Spielberger et al. 1970), and ratings of a series of 33 symptoms that have been shown in previous studies to be sensitive to benzodiazepine withdrawal (benzodiazepine withdrawal scale; e.g., Griffiths et al. 1993; Mintzer et al. 1999). Observer ratings Ratings of motor restlessness, sedation, perspiration, anxiety, confusion, hypervigilance, irritable mood, resting tremor and tense were made by a research assistant based on the his/her observations and global impression of the participant’s behavior, including any verbal comments made by the participant since the last rating. Physiological measures Blood pressure, heart rate and skin temperature were monitored automatically by a Criticare non-invasive patient monitor (Criticare Systems, Inc., Waukesha, WI, USA), with data output measured minute by minute. Heart rate was measured via electrocardiogram leads. The blood pressure cuff was placed on the non-dominant arm. Skin temperature was monitored using a skin-surface thermistor (Yellow Springs Instrument Co., Yellow Springs, OH) taped to the index finger of the non-dominant hand. Respirations were monitored visually by the research assistant at 5-min intervals. For the measures that were recorded minute to minute, data were averaged and analyzed as 3-min blocks. Data analysis Data from participant and observer ratings were analyzed as difference scores (i.e., change from pre-injection values; calculated as change from the −10-min baseline time point; positive scores indicate an increase relative to baseline), with the exception of visual analog scale ratings of strength of drug effect, bad effects, and good effects that were analyzed as raw scores (because all baseline scores were 0). Inspection of time-course functions for participant and observer ratings indicated that peak effects (for all measures that showed effects of flumazenil) occurred at the time point immediately after completion of the
injection and typically dissipated before the 20-min assessment. Data from the “post-session” retrospective assessment were comparable to the peak-effects data. Therefore, only peak effects (or peak change from baseline, for measures analyzed as difference scores) will be presented for participant and observer ratings. Inspection of time-course functions for physiological measures indicated that for measures that showed effects of flumazenil, the effects were sustained throughout the approximate 60-min session. Therefore, physiological data were analyzed as area under the time-course curve. Peak effects for participant and observer ratings (or peak change from baseline) and area under the curve for physiological measures were analyzed by means of a mixed-design analysis of variance (ANOVA) with group (diazepam and placebo) and day (i.e., days of capsule ingestion; 0, 1, 7, 14, 28) as factors; Huynh–Feldt corrections for compound symmetry were applied. Significant interactions were followed-up with analyses of simple effects (Keppel 1991). All analyses were considered statistically significant when P≤0.05. Data from the serum samples and the 2-week follow-up period were not analyzed statistically.
Results Serum levels (diazepam group only) Due to technical difficulties, serum samples from two participants were not available for analysis for day 28. Although data from these participants were excluded from the means reported below, their available data followed a pattern similar to that of the other participants. As expected, diazepam levels increased the first week and then leveled off to a steady-state for the remainder of the study (mean diazepam levels for days 1, 7, 14, and 28 were 119, 379, 376, and 383 ng/ml, respectively; n=8). Peak diazepam levels were within the therapeutic range for anti-anxiety effects (300–400 ng/ml; Greenblatt et al. 1981). By day 7, levels of nordiazepam were higher than diazepam levels and continued to increase throughout the 28 days of diazepam exposure (mean nordiazepam levels for days 1, 7, 14, and 28 were 26, 385, 513, and 583 ng/ ml, respectively; n=8). This differential pattern of accumulation is consistent with the slower elimination of nordiazepam [t1/2=60 h (approximately)] relative to diazepam (t1/2=24–48 h). Participant ratings Visual analog scale ratings of strength of drug effect and bad drug effects were increased in the diazepam group relative to placebo after 7 days of nightly capsule ingestion, and this increase was maintained after 14 days and 28 days (Fig. 1). The increase was reflected in a significant group×day interaction (F4,64=14.00 for ratings of strength of drug effect; F4,64=8.08 for ratings of bad drug effects). Simple-effects analyses indicated that ratings
263 Fig. 1 Mean ratings on visual analog scales strength of drug effect, bad drug effects, and drug liking for the diazepam (circles; n=10) and placebo (squares; n=8) groups during each flumazenil session. Ratings were made on a 100-point line marked from 0 (not at all) to 100 (extremely) for strength of drug effect and bad drug effects, and marked 0 (dislike), 50 (neutral), and 100 (like) for drug liking. Yaxes show peak ratings (expressed as change from preinjection baseline for drug liking). X-axes show days of nightly capsule ingestion, with 0 indicating the day before the first nightly capsule ingestion. Brackets indicate ±1SEM; for clarity of presentation, either the top or lower bracket is sometimes omitted. An asterisk indicates a significant difference between the diazepam and placebo groups on that day (P≤0.05, simple effects tests)
Fig. 2 Mean composite scores on the STAI and the benzodiazepine withdrawal scale for the diazepam (circles; n=10) and placebo (squares; n=8) groups during each flumazenil session. Y-axes show peak scores expressed as change from preinjection baseline. Other details as in Fig. 1
were significantly increased in the diazepam group relative to placebo on days 7, 14, and 28. A somewhat different pattern was evident on visual analog ratings of drug liking, which continued to decrease through 14 days of diazepam exposure and then stabilized (Fig. 1). For drug liking, there was a significant group×day interaction (F4,64=3.92) and simple-effects analyses indicated that drug-liking ratings were significantly increased in the diazepam group relative to placebo on day 14. The diazepam and placebo groups did not differ in ratings of good drug effects after flumazenil, with both groups indicating a lack of good drug effects associated with flumazenil throughout the study (data not shown). Similar to the pattern of effects on visual analog ratings of strength of drug effect and bad effects, the STAI composite score (anxiety; Fig. 2) was elevated after 7 days of nightly diazepam ingestion, and stayed elevated,
relative to the placebo group after 14 days and 28 days of diazepam (significant group×day interaction, F4,64=2.97). Simple-effects analyses revealed that the STAI composite score was significantly increased in the diazepam group relative to placebo on day 14. Unlike the pattern with visual analog ratings, simple-effects analyses also revealed a significant difference between groups before capsule ingestion began (day 0), such that the STAI composite score was significantly lower in the diazepam group relative to placebo, although the groups did not differ after 1 day of capsule ingestion. Consistent with the typical pattern for participant ratings, the composite score on the benzodiazepine withdrawal scale (Fig. 2) was also elevated after 7 days of nightly diazepam ingestion, and stayed elevated, relative to the placebo group after 14 days and 28 days (significant group×day interaction: F4,64=7.40). Simple-
264
effects analyses indicated that the composite score was significantly increased in the diazepam group relative to placebo on days 7, 14, and 28. The individual items of the withdrawal scale were analyzed, and several showed a significant main effect of group (blurred vision, heart pounding, pins and needles, irritable, things moving) and/ or a significant group×day interaction (blurred vision, pins and needles, things moving, confused, dizzy, difficulty concentrating). For items with a significant group×day interaction, simple-effects analyses indicated that scores were significantly increased in the diazepam group relative to placebo on day 7 (pins and needles, and dizzy), day 14 (blurred vision, dizzy, and things moving), and day 28 (blurred vision, pins and needles, dizzy, and things moving). Observer ratings There was a significant main effect of group on observer ratings of anxiety such that ratings were increased in the diazepam group relative to placebo (F1,16=4.55; data not shown). There were no significant main effects of group on any other observer ratings, nor were there significant group×day interactions on any items.
Fig. 3 Mean blood pressure and skin temperature values for the diazepam (circles; n=9) and placebo (squares; n=7) groups during each flumazenil session. Y-axes show area under the time-course curve. Other details as in Fig. 1
Physiological measures Due to technical difficulties during data collection, physiological data were not available for two participants (one each in the diazepam and placebo groups) from one session each. Physiological data from these two participants were excluded from the analyses. Similar to the pattern for participant ratings, blood pressure (Fig. 3) increased after 7 days of nightly diazepam ingestion, and stayed elevated, relative to the placebo group after 14 days and 28 days, as reflected in a significant group×day interaction (F4,56=4.21 for systolic; F4,56=5.43 for diastolic). Simple-effects analyses indicated that blood pressure (both systolic and diastolic) was significantly increased in the diazepam group relative to placebo on days 7, 14, and 28. For skin temperature (Fig. 3), there was a significant group×day interaction (F4,56=3.65), but simple-effects analyses did not indicate significant differences between groups on any day. This is likely due to the fact that, while there are downward trends in temperature as a function of day in the diazepam group, these did not contribute to large overall differences relative to placebo in any mean day temperature. There were no significant main effects of group or group×day interactions on heart rate or respiration rate (data not shown).
265
Two-week follow-up During follow-up, three participants in the diazepam group and none in the placebo group reported symptoms on a standardized measure of clinical benzodiazepine withdrawal (Busto et al. 1989). For these three participants, reported symptoms included: restlessness (n=3), sleeplessness (n=2), tense (n=2), fatigue (n=2), sweating (n=1), achy (n=1), anxious (n=1), stiff (n=1), weak (n=2), and loss of appetite (n=2). Scores on the clinical benzodiazepine withdrawal scale (Busto et al. 1989) for these participants were highest on the first day they were interviewed post-study (day 1 post-study for two of the participants; day 2 post-study for one participant who was not interviewed on day 1 but was interviewed on day 2). On the first interview day post-study, the mean score was 9.3 out of a maximum possible score of 80, indicating mild symptomology. In most cases, the symptoms were reported as mild; in the remaining cases, the symptoms were reported as moderate. No severe symptoms were reported and no interventions were needed. On the first day of interview post-study, two of the three participants reported decreased total amounts of sleep for the preceding night, but none reported increased latency to fall asleep. By the end of the 2-week follow-up period, no participants reported symptoms that appeared to be related to study participation.
Discussion Results of the present study support previous research indicating that flumazenil-precipitated withdrawal occurs in human participants after short-term exposure to benzodiazepines (Griffiths et al. 1993). Flumazenil administered prior to diazepam ingestion, and after 1 day of diazepam, did not produce changes in participant ratings. In contrast, after 7, 14, and 28 days of diazepam exposure, flumazenil administration produced effects consistent with benzodiazepine withdrawal in the diazepam group relative to placebo. Flumazenil produced increases in visual analog ratings of strength of drug effect and bad drug effects, and decreases in drug-liking ratings. In addition, flumazenil produced increases on both the benzodiazepine withdrawal scale and the STAI, a measure of anxiety. Consistent with results of previous human laboratory studies, flumazenil’s effects on participant ratings peaked immediately after completion of the 5min injection and rapidly dissipated thereafter (Brogden and Goa 1991; Frey et al. 1996; Griffiths et al. 1993; Mintzer et al. 1999). Flumazenil also produced modest elevations in blood pressure and decreases in skin temperature that were sustained throughout the approximate 60-min session. The finding of increased blood pressure is consistent with results of prior studies assessing the effects of flumazenil in therapeutic users of benzodiazepines (Harrison-Read et al. 1996; Mintzer et al. 1999), although inconsistent with results of one study of precipitated withdrawal in healthy volunteers that failed
to show significant physiological effects (Griffiths et al. 1993). The absence of changes in the placebo group in response to flumazenil is consistent with previous demonstrations that the direct effects of flumazenil in healthy volunteers are minimal (Griffiths et al. 1993; Mintzer et al. 1999). With respect to the effect of duration of agonist exposure on flumazenil-precipitated withdrawal, results of the present study indicate that precipitated withdrawal occurred after 7 days, but not 1 day, of diazepam exposure, and that the intensity of withdrawal effects was generally of similar magnitude after 7, 14, and 28 days (except for effects on drug-liking ratings, which appeared to be somewhat more intense after 14 days than after 7 days). The suggestion that precipitated withdrawal does not occur after 1 day of agonist exposure is generally consistent with results of previous human laboratory (Griffiths et al. 1993) and preclinical (Lukas and Griffiths 1984; Wilson and Gallager 1988; but cf. Rosenberg and Chiu 1985 who showed some precipitated withdrawal effects in cats after 1 day of flurazepam exposure) studies. Likewise, the finding of precipitated withdrawal effects after 7 days of agonist exposure is consistent with results of previous human (Griffiths et al. 1993) and preclinical (Rosenberg and Chiu 1985; Lukas and Griffiths 1982, 1984; Wilson and Gallager 1988) studies. The finding of precipitated withdrawal effects after only 7 days of exposure suggests that the physiological processes underlying dependence are already at play after a relatively brief exposure duration. Although the clinical implications of this finding are uncertain, they suggest that prudence should be exercised in prescribing benzodiazepines even on a short term basis. The finding that precipitated withdrawal effects were of similar magnitude after 7, 14, and 28 days is consistent with the results of one preclinical study that indicated that the intensity of precipitated withdrawal was nearly maximal after 7 days of exposure on most variables and did not increase further after 35 days or 70 days (Rosenberg and Chiu 1985). In addition, another preclinical study indicated that the intensity of precipitated withdrawal was maximal after 2 weeks of exposure and did not increase further after 4 weeks of exposure (Kaminski et al. 2003; although in that study, precipitated withdrawal was not assessed after only 7 days of exposure). However, the absence of further increases in intensity as agonist exposure was increased beyond 7 days appears inconsistent with results of other preclinical studies, suggesting that the intensity and/or frequency of precipitated withdrawal effects increases further as a function of agonist exposure duration for up to 35 days of agonist exposure (Lukas and Griffiths 1982, 1984; Wilson and Gallager 1988; Zharkovskii and Zharkovskaia 1987 as cited in Woods et al. 1992). It is important to note that comparisons between human and preclinical studies are somewhat complicated by the following factors. First, dosing parameters often differ in preclinical versus human studies, with many preclinical studies evaluating doses and schedules of dosing that far
266
exceed those that are clinically relevant in humans (Woods et al. 1987). Second, there appears to be some inconsistency in effects among different preclinical studies (as described above) as well as among different measures of withdrawal even within single preclinical studies (Lukas and Griffiths 1984). As we noted earlier, in a prior human study in our laboratory, using a group design to manipulate duration of lorazepam exposure, precipitated withdrawal effects were not greater in participants pretreated for 14 days than in participants treated for 7 days, consistent with results of the present study. However, in that prior study, effects were actually somewhat attenuated after 14 days relative to 7 days, whereas in the present study, there was simply no change in intensity. As discussed earlier, the attenuation observed in the prior study may have resulted from uncontrolled for group factors, and this concern was addressed in the present study using a within-subject design. Taken together, results of Griffiths et al. (1993) and the present study suggest that precipitated withdrawal does not increase in intensity as a function of duration of agonist exposure beyond 7 days of exposure when doses in or near the therapeutic range are assessed in healthy volunteers. Interestingly, studies in chronic daily users of therapeutic doses of benzodiazepines suggest that flumazenil-precipitated withdrawal also does not increase as a function of agonist exposure duration when exposure is on the order of years rather than weeks. For example, in a recent study that assessed flumazenil-precipitated withdrawal in patients who had used low doses of benzodiazepines daily for 1–15 years (Mintzer et al. 1999), the intensity of symptoms precipitated by flumazenil was correlated with the magnitude of the maintenance dose, but not the duration of exposure (as measured by duration in months of continual exposure as well as by total cumulative exposure to benzodiazepines, calculated as average total daily dose in diazepam equivalent multiplied by total days of use; Petursson and Lader 1984). Likewise, in two studies in chronic users of low therapeutic doses of diazepam (for 5–15 years: Bernik et al. 1991; for 5–28 years: Bernik et al. 1998), the intensity of precipitated withdrawal was correlated with a history of panic attacks, but not with the duration of agonist exposure (as measured by total cumulative benzodiazepine exposure). During the 2-week follow-up period, only three of the ten participants in the diazepam group reported symptoms of withdrawal, and these participants reported only mild symptoms (e.g., slight anxiety and sleeplessness), suggesting that flumazenil precipitated more severe withdrawal than that which occurred during subsequent “spontaneous” withdrawal. Results of preclinical studies that have assessed both precipitated and spontaneous withdrawal are inconsistent, with some studies indicating greater severity when flumazenil is used to precipitate withdrawal (Kaminski et al. 2003; Lukas and Griffiths 1982; Rosenberg and Chiu 1985) and others indicating greater severity with spontaneous withdrawal (McNicholas and Martin 1982; Scherkl and Frey 1986). Experimental studies with human participants do not typically include
spontaneous withdrawal controls or a spontaneous withdrawal phase. While results of the present study suggest that the intensity of withdrawal precipitated by flumazenil is greater than that of spontaneous withdrawal, it is important to note that given that spontaneous withdrawal effects were examined at the end of the study after participants had been exposed to multiple flumazenil challenges, it is possible that flumazenil administration itself decreased the degree of physical dependence, resulting in reduced spontaneous withdrawal symptoms. The notion that flumazenil may alter the course of physical dependence is supported by results of preclinical studies indicating that repeated flumazenil administration reduces the intensity and/or frequency of precipitated withdrawal effects (Gallager et al. 1986; Lamb and Griffiths 1985). Likewise, results of studies in chronic benzodiazepine users who have discontinued benzodiazepine use suggest that flumazenil administration reduces spontaneous withdrawal symptoms when given 12 h (Gerra et al. 1993) or even several weeks or years (Lader and Morton 1992; Saxon et al. 1997) after discontinuation. One theory is that chronic exposure to benzodiazepines is associated with a shift in the benzodiazepine receptor set-point toward inverse agonism, causing antagonists to act as inverse agonists, and that flumazenil resets the receptor’s set-point (Nutt 1990; Nutt et al. 1990). However, it should be noted that the notion that flumazenil resets the receptor’s set-point appears inconsistent with findings that repeated flumazenil administration does not restore sensitivity to some of the sedative effects of benzodiazepines after tolerance has developed to these effects with chronic use (Lamb and Griffiths 1985; Sannerud et al. 1989). Thus, while the potential for therapeutic use of flumazenil in patients attempting to discontinue chronic benzodiazepine use is intriguing (Savic et al. 1991), further research is needed to understand the underlying mechanisms as well as to test the reliability of the effects. In summary, the results of this study indicate that flumazenil (1 mg/70 kg) precipitated withdrawal after 7 days but not 1 day of diazepam exposure. Symptoms of precipitated withdrawal were similar to those previously reported and were consistent with withdrawal symptoms that can occur during spontaneous withdrawal from chronic benzodiazepine use. The symptoms were of similar magnitude after 7, 14, and 28 days of diazepam exposure, suggesting that duration of exposure, at least under the time frame assessed in this study, did not influence the intensity of withdrawal beyond the first week of exposure. Acknowledgments This project was supported by National Institute on Drug Abuse Research Grant DA-03889. The authors thank Ryan Vandrey for technical assistance, Tim Mudric and Paul Nuzzo for statistical consultation, John Yingling for programming and technical support, and David Ginn, M.D. for medical coverage.
267
References Bernik MA, Gorenstein C, Gentil V (1991) Flumazenil-precipitated withdrawal symptoms in chronic users of therapeutic doses of diazepam. J Psychopharmacol 5:215–219 Bernik MA, Gorenstein C, Vieira Filho AH (1998) Stressful reactions and panic attacks induced by flumazenil in chronic benzodiazepine users. J Psychopharmacol 12:146–150 Brogden RN, Goa KL (1991) Flumazenil. A reappraisal of its pharmacodynamic properties and therapeutic efficacy as a benzodiazepine antagonist. Drugs 42:1061–1089 Bruce SE, Vasile RG, Goisman RM, Salzman C, Spencer M, Machan JT, Keller MB (2003) Are benzodiazepines still the medication of choice for patients with panic disorder with or without agoraphobia? Am J Psychiatry 160:1432–1438 Busto UE, Sykora K, Sellers, EM (1989) A clinical scale to assess benzodiazepine withdrawal. J Clin Psychopharmacol 9:412– 416 Frey JM, Rush CR, Griffiths RR (1996) Physiological, behavioral and subjective effects of flumazenil-precipitated withdrawal in benzodiazepine-dependent methadone clients. Poster presented at the annual meeting of the Society for Neuroscience, Washington, DC Gallager DW, Heninger K, Heninger G (1986) Periodic benzodiazepine antagonist administration prevents benzodiazepine withdrawal symptoms in primates. Eur J Pharmacol 132:31–38 Gerra G, Marcato A, Caccavari R, Fertonani-Affini G, Fontanesi B, Zaimovic A, Avanzini P, Delsignore R (1993) Effectiveness of flumazenil (RO 15-1788) in the treatment of benzodiazepine withdrawal. Curr Ther Res Clin Exp 54:580–587 Gerra G, Giucasto G, Zaimovic A, Fertonani G, Chittolini B, Avanzini P, Caccavari R, Delsignore R (1996) Intravenous flumazenil following prolonged exposure to lormetazepam in humans: lack of precipitated withdrawal. Int Clin Psychopharmacol 11:81–88 Greenblatt DJ, Laughren TP, Allen MD, Harmatz JS, Shader RI (1981) Plasma diazepam and desmethyldiazepam concentrations during long-term diazepam therapy. Br J Clin Pharmacol 1:35–40 Griffiths RR, Weerts EM (1997) Benzodiazepine self-administration in humans and laboratory animals-implications for problems of long-term use and abuse. Psychopharmacology (Berl) 134:1–37 Griffiths RR, Evans SM, Guarino JJ, Roache JD, Furman WR, Liebson I, Schwam EM (1993) Intravenous flumazenil following acute and repeated exposure to lorazepam in healthy volunteers: antagonism and precipitated withdrawal. J Pharmacol Exp Ther 265:1163–1174 Harrison-Read PE, Tyrer P, Lawson C, Lack S, Fernandes C, File SE (1996) Flumazenil-precipitated panic and dysphoria in patients dependent on benzodiazepines: a possible aid to abstinence. J Psychopharmacol 10:89–97 IMS Health (2003) Available from http://www.imshealth.com/ims/ portal/front/articleC/O,2777,6599_42720951_43989693,00. html, accessed 2003 Kaminski BJ, Sannerud CA, Weerts EM, Lamb RJ, Griffiths RR (2003) Physical dependence in baboons chronically treated with low and high doses of diazepam. Behav Pharmacol 14:331–342 Kaufman DW, Kelly JP, Rosenberg L, Anderson TE, Mitchell AA (2002) Recent patterns of medication use in the ambulatory adult population of the United States: the Slone survey. JAMA 287:337–344 Keppel G (1991) Design and analysis: a researcher’s handbook, 3rd edn. Prentice Hall, Englewood Cliffs Lader MH, Morton SV (1992) A pilot study of the effects of flumazenil on symptoms persisting after benzodiazepine withdrawal. J Psychopharmacol 6:357–363 Lamb RJ, Griffiths RR (1985) Effects of repeated Ro 15-1788 administration in benzodiazepine-dependent baboons. Eur J Pharmacol 110:257–261
Lamb RJ, Griffiths RR (1987) Effects of Ro 15-1788 and CGS 8216 in diazepam-dependent baboons. Eur J Pharmacol 143:205–212 Lukas SE, Griffiths RR (1982) Precipitated withdrawal by a benzodiazepine receptor antagonist (Ro 15-1788) after 7 days of diazepam. Science 217:1161–1163 Lukas SE, Griffiths RR (1984) Precipitated diazepam withdrawal in baboons: effects of dose and duration of exposure. Eur J Pharmacol 100:163–171 Martin WR, Sloan JW, Wala E (1990) Precipitated abstinence in orally dosed benzodiazepine-dependent dogs. J Pharmacol Exp Ther 255:744–755 McNicholas LF, Martin WR (1982) The effect of a benzodiazepine antagonist, RO15-1788, in diazepam dependent rats. Life Sci 31:731–737 McNicholas LF, Martin WR, Sloan JW, Wala E (1988) Precipitation of abstinence in nordiazepam- and diazepam-dependent dogs. J Pharmacol Exp Ther 245:221–224 Mintzer MZ, Stoller KB, Griffiths RR (1999) A controlled study of flumazenil-precipitated withdrawal in chronic low-dose benzodiazepine users. Psychopharmacology 147:200–209 Nutt DJ (1990) Pharmacological mechanisms of benzodiazepine withdrawal. J Psychiatr Res 24(Suppl 2):105–110 Nutt DJ, Glue P, Lawson C, Wilson S (1990) Flumazenil provocation of panic attacks. Arch Gen Psychiatry 47:917–925 Petursson H, Lader M (1984) Dependence on tranquilizers, Maudsley monographs 28. Oxford University Press, Oxford Physicians’ Desk Reference (2003) 57th edn. Medical Economics, Montvale Rosenberg HC, Chiu TH (1985) Time course for development of benzodiazepine tolerance and physical dependence. Neurosci Biobehav Rev 9:123–131 Sannerud CA, Cook JM, Griffiths RR (1989) Behavioral differentiation of benzodiazepine ligands after repeated administration in baboons. Eur J Pharmacol 167:333–343 Sannerud CA, Allen M, Cook JM, Griffiths RR (1991) Behavioral effects of benzodiazepine ligands in non-dependent, diazepamdependent and diazepam-withdrawn baboons. Eur J Pharmacol 202:159–169 Savic I, Widen L, Stone-Elander S (1991) Feasibility of reversing benzodiazepine tolerance with flumazenil. Lancet 337:133–137 Saxon L, Hjemdahl P, Hiltunen AJ, Borg S (1997) Effects of flumazenil in the treatment of benzodiazepine withdrawal—a double-blind pilot study. Psychopharmacology (Berl) 131:153– 160 Scherkl R, Frey HH (1986) Physical dependence on clonazepam in dogs. Pharmacology 32:18–24 Sloan JW, Martin WR, Wala E (1991a) Duration of diazepam (DZ) treatment and the intensity of flumazenil precipitated abstinence (FPA) in dogs. FASEB J 5:A703 Sloan JW, Martin WR, Wala E, Dickey KM (1991b) Chronic administration of and dependence on halazepam, diazepam, and nordiazepam in the dog. Drug Alcohol Depend 28:249–264 Sloan JW, Martin WR, Wala E (1993) Effect of the chronic dose of diazepam on the intensity and characteristics of the precipitated abstinence syndrome in the dog. J Pharmacol Exp Ther 265:1152–1162 Spielberger CD, Gorsuch RL, Lushene RE (1970) STAI manual for the state-trait anxiety inventory (“Self-Evaluation Questionnaire”). Consulting Psychologists, Palo Alto Wilson MA, Gallager DW (1988) Ro 15-1788-induced seizures in rats continually exposed to diazepam for prolonged periods. Epilepsy Res 2:14–19 Woods JH, Katz JL, Winger G (1987) Abuse liability of benzodiazepines. Pharmacol Rev 39:251–419 Woods JH, Katz JL, Winger G (1992) Benzodiazepines: use, abuse, and consequences. Pharmacol Rev 44:151–347 Zharkovskii AM, Zharkovskaia TA (1987) Characteristics of the abstinence syndrome induced by the administration of the benzodiazepine antagonist CGS 8216 to rats following chronic treatment with diazepam. Biull Eksp Biol Med 104:318–321
