Absence of emergent suicidal ideation during treatment: A comparative, controlled, double-blind analysis employing several distinct antidepressants

Absence of emergent suicidal ideation during treatment: A comparative, controlled, double-blind analysis employing several distinct antidepressants
of 7
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  DEPRESSION :73 79 z 1994) ABSENCE OF EMERGENT SUICIDAL IDEATION DURING BLIND ANALYSIS EMPLOYING SEVERAL DISTINCT ANTIDEPRESSANTS TREATMENT': A COMPARATn/TE, CONTROLLED, DOUBLE- Gary D. Tollefson, M.D., Ph.D., Sherrie L. Tollefson, R.N., M.B.A., Mary E. Sayler, M.S., and Michael G. Luxenberg, Ph.D. zyxw We investigated 243 patients fiom a series of comparator and/or placebo-con- trolled, double-blind trials of several mechanistically distinct antidepressants. Data were analyzed for changes in suicidal ideation during trial participation. Ambulato referrals with a current DSM-III-R major depression, corrobo- rated by a structured interview, were screened; those fidjilling entrance crite- ria began a 1-week, single-blind placebo run-in. Pool 1 (placebo comparator) subjects who continued to meet entrance criteria at their baseline visit were blindly randomized to receive either placebo (n zyxw   49) or one of several novel antidepressant drugs (n = 81). Pool 2 subjects (active comparator) were ran- domized to either a novel antidepressant (n = 66 or a 2ricyclic antidepressant comparator (n = 57). Trial duration rangedfiom zyx   to 8 weeks. Patient compli- ance was monitored by a regular pill countfiom each subject's blister pack. Analyses were conducted on Pool 1 and Pool 2 separately. Ego-dystonic sui- cidal preoccupation or acts were not reported in either group. There was a statistically significant treatment diflerence in Hamilton Depression Rating Scale item 3 mean peak change (Pool 1); patients treated with a novel antide- pressant demonstrated a mean decrease in HDRS item 3 (improvement), those patients randomized to placebo experienced a mean score increase (worsening). No statistically signijicant treatment diflerence in mean peak change difler- ence factors (CDFs) between the novel antidepressants and conventional tricyclic antidepressant comparators (Pool 2) was observed. A statistically sig- nifacant cowelation between item 3 change and total HDRS-17 change also emerged (hoth treatments within each pool). Increased suicidal ideashehaviors may emerge during depression 'b-eatment. Howevel; in this dy t occurred moreji-equently among placebo recipients, sug- gesting a change in suicidal ideation was a reflection of the underlying disease pcess. Not surprisingly, antidepressants improved suicidal ideation to a greater extent than placebo. Depression 2:73-79 (1994). @ 994 W~V-L~SS BC. Key words: antidepressants, (emergent) suicidality, depression, serotonin, norepinephrine Maior INTRODUCTION depression (MD) is associated with a high rate of suicidal ideation, behaviors, or acts.' The emer- gence or intensification of suicidal ideation during the pharmacologic treatment of MD has been reported.* However, a causal link to drug therapy has not been established. Fava and Rosenbaum3 suggested that treatment-emergent ideation can be attributed to many non-drug factors such as the natural history of MD, the disillusionment accompanying non-response, zyxwv Department of Psychiatry at St. Paul-Ramsey Medical Cen- ter, St. Paul G.D.T., S.L.T.), and Professional Data Analysts, Minneapolis M.G.L.), Minnesota; Lilly Research Laborato- ries, Eli Lilly and Company, Indianapolis, Indiana G.D.T., S.L.T., M.E.S.). Received for publication February 15 1994; revised August 4 1994; accepted September 26 1994. Address reprint requests to Gary D. Tollefson M.D. Ph.D. Execu- tive Director Lilly Research Laboratories Eli Lilly and Company Lilly Corporate Center Drop Code 2128 Indianapolis IN 46285. 994 WILEY-LISS INC  74 zyxwvus ollefson et al. zyxw or an early psychomotor mobilization that presages mood and cognitive recovery. Fawcett emphasized an increased risk of suicide in the depressed patient was seen in the presence of comorbid panic, psychic anxi- ety, turmoil, or moderate alcohol abuse.4 Additional considerations include an increase in the depression prior to a therapeutic response, non-compliance, fail- ure to achieve a therapeutic drug level or effect, psy- chosocial disruptions, or concurrent medical illness. Any of these coincident factors could mediate suicidality during the course of a therapeutic intervention. Alternatively, several authors have postulated that an antidepressant-associated increase in suicidality could be mediated via its neurochemical profile, e.g., selec- tive uptake inhibition of norepinephrine’ or seroto- nin.6 Such a mechanism has been alleged to underlie anecdotal reports of intense, violent suicidal preoccu- pation.“” Subsequently, indirect explanations, e.g., adverse drug reactions (more common to certain classes of antidepressants) have been invoked to ex- plain the temporal association of suicide during drug therapy.”-” zyxwvutsr   scientific approach to address causality is through controlled, blinded clinical trials. The benefit of including a placebo comparison is the ability to ad- dress a base rate of a phenomenon such as suicidality. In this investigation, we tested the hypothesis that treatment-emergent suicidal ideation, as part of the depressive syndrome, would be seen at least as fre- quently with placebo as with a spectrum of antidepres- sant therapies. A secondary study question was, within a diverse mechanistic group of novel and conventional antidepressants, whether evidence suggested that one mechanistic class manifested a higher incidence of emergent ideation than another during pharmaco- therapy. To answer these two questions, we analyzed data from a series of double-blind, controlled trials involving several chemically distinct antidepressants compared to either placebo (Pool 1) or a conventional tricyclic antidepressant (TCA) (Pool 2). Individual and group variation in the third item (suicide) from the Hamilton Depression Rating Scale (HDRS),16 spontaneous event reporting, and the clinical course were evaluated from this series of acute pharamcotherapy trials. METHODS CLINICAL TRIALS A series of parallel double-blind, placebo- or comparator-controlled trials investigating several novel antidepressant compounds were carried out in an ambulatory depressed population. Subjects from age 18 to 64 (excluding a single geriatric protocol en- rolling patients age 63 to 80) experiencing first or re- current episode of unipolar major depression (MD) without psychotic features corroborated by a Struc- tured Clinical Interview for Diagnosis” and congru- ent with DSM-III-RI8 criteria were sought. The primary outcome measure was the HDRS. A z ini- mum HDRS17 otal score of 18 was required at screen- ing and again 1 week later after a single-blind placebo run-in. In order to be randomized the HDRS score also could not have fallen in excess of 25% from its corresponding screening value. Subjects were excluded if they had a recent attempt or active suicidal planning at screening, a severe pre- existing medical condition that contraindicated drug therapy, and/or a clinically important abnormality on screening laboratories. Additional exclusions were a positive urine drug screen, active alcohol or drug abuse/dependency within the previous 6 months, or any other active Axis I diagnosis. Other CNS active drugs were not permitted within 2 weeks of screening or 4 weeks if the drug was fluoxetine, a monoamine oxidase inhibitor, or other investigational compound. All participants provided a written informed consent for study participation. A description of each trial follows: Etoperidone. Etoperidone is a non-tricyclic mol- ecule that is chemically related to the triazolopyridines (e.g., trazadone).’’ It is a serotonin (5-HT) uptake in- hibitor, also active at post-synaptic 5-HT2 eceptors. Etoperidone additionally exhibits alpha-adrenergic blockade. Its principle metabolite is a meta-chloro- phenyl-piperazine (MCPP). Protocol EC: Patients assigned to etoperidone were titrated up to 300 mg/day (n zy   9) or alterna- tively received placebo (n = 9) for 6 weeks follow- ing baseline randomization (Pool 1). Protocol EGO: Individuals, aged 63-80 years, were randomized to either etoperidone (50-300 mg per day) n = 11) as tolerated or placebo (n = 9) and followed for 8 weeks (Pool 1). Protocol ETI: Patients were randomly assigned (ratio 3 : 1) to etoperidone (n = 17) up to 400 mg/ day as tolerated or imipramine up to 200 mg/day (n = 6). Treatment, under double-blind, was for 6 weeks following randomization (Pool 2). Fluparoxan. Fluparoxan” is a highly selective al- pha-2 adrenoceptor antagonist comparable to yohim- bine or idazosam. Its selective attenuation of the adrenergic autoreceptor presumably increases net available synaptic norepinephrine WE). Fluparoxan recipients (n = 11) received 8 mg twice daily while placebo recipients (n = 12) received an equal number of inert capsules for a total of 3 5 days (Pool 1). Nefazodone. Nefazodone is a selective 5-HT anti- depressant which inhibits 3H-spiperone binding.” Chronic administration has been associated with down-regulation of 5-HT2 binding sites. Nefazodone exerts negligible affinity at cholinergic, adrenergic, histaminergic, or dopaminergic receptors. Patients in this protocol were assigned to one of four fixed dosages (n = 40) of nefazodone (25 mg b.i.d., 50 mg b.i.d., 100 mg b.i.d., 150 mg b.i.d.) or  Research Article: Antidepressant-Related Change in Suicidality zyx 5 z placebo (n zyxwvutsr   10) at their baseline visit and were fol- lowed for 6 weeks (Pool 1). Adinazolam. Adinazolam mesylate is a triazaloben- zodiazepine reportedly effective in several pre-clinical animal models and outpatient clinical depression trials.22 Adinazolam is active at the benzodiazepine-gamma- amino butyric acid-chloride supramolecular complex and also reportedly is an NE uptake inhibitor leading to a down-regulation of beta-NE binding sites.23 Subjects were randomly assigned to either a combi- nation of the tricyclic desipramine SO mg plus a pla- cebo capsule (n = 20 or desipramine SO mg plus adinazolam 10 mg (n = 20). By investigator discretion, dosage could be titrated to a maximum of two cap- sules, three times daily as tolerated. The trial contin- ued for a total of 6 weeks (Pool 2). Dothiepin. Dothiepin is a novel tricyclic antide- pressant with a chemical structure similar to amitrip- tyline.*’ It preferentially inhibits the uptake of NE, down-regulates cortical beta-NE receptors, and re- duces NE-induced cyclic AMP formation.*‘ Following the 1 week screen, subjects received dothiepin 50 mg (n = lo), doxepin 50 mg (n = lo), or placebo (n = 9) at night for 3 days; dosages were then titrated over 4 to 7 days to achieve a minimum three capsules for zyxwvutsr   weeks (Pool 1, dothiepin vs. placebo; Pool 2, dothiepin vs. doxepin). Fluoxetine. Fluoxetine is a selective 5-HT uptake inhibitor that enhances synaptic -HT through inhibi- tion of presynaptic 5-HT autoregulatory sites.25 After a 1-week placebo run-in, eligible subjects were randomized to either fluoxetine 20 mg in the morning (evening placebo) (n = 19) or imipramine SO mg in the evening (AM placebo) (n = 21). Dosages were titrated by clinical indication and/or patient tolerance to a maximum dosage of fluoxetine 60 mg or imipramine 30 mg. Minimum dosages were 20 and 150 mg, re- spectively. Both groups were followed for 8 weeks af- ter randomization (Pool 2). ANALYTICAL AND STATISTICAL METHODS For these analyses, the clinical studies were divided into Pool 1 (novel antidepressants vs. placebo) and Pool 2 (novel antidepressants vs. tricyclic antidepres- sants). Note that one clinical study compared a novel antidepressant (dothiepin) with both placebo and a tri- cyclic antidepressant so the dothiepin arm is included in both pools. Within pools, several analyses were performed on the HDRS item 3 (suicidal ideation) score. First, the mean peak change difference factor (CDF) was com- pared between treatments. The peak CDF was defined as the maximum HDRS item 3 score during the double-blind portion of the trial minus the baseline score (the measurement prior to initiation of therapy after the single-blind placebo run-in portion). Possible CDFs were: 0 = no change in suicidal ideation during the investigation; a negative score (<O) which indicated only improvement (or a reduction) in ideation during treatment; or a positive (>O) which indicated a worsen- ing (or increase) in suicidal ideation anytime during the trial independent of final outcome. Thus, CDFs captured the highest post-randomization HDRS item 3 score regardless of study week, score duration, or fi- nal endpoint visit. Second, the relative proportion of patients at each peak CDF score were compared between treatments. Third, the proportion of patients who improved (peak CDF < 0), experienced no change (peak CDF = 0), or worsened (peak CDF > 0) from their baseline score were compared between treatments. Fourth, the pro- portion of patients whose HDRS item 3 score in- creased from 0 or 1 at baseline to 3 or 4 at any point during the double-blind portion of the trial were com- pared between patients (emergence of substantial sui- cidal ideation). Fifth, mean change in HDRS total score and item 3 from baseline to endpoint (last mea- surement during the double-blind portion of the trial) were compared between treatments. The relationship between change in HDRS item 3 and the HDRS 17 item total from baseline to last observation were also determined. In addition, age, gender distribution, baseline HDRS total score, and baseline HDRS item 3 were compared between treatments within pools. All CDF analyses involving HDRS item 3 included all patients zyx it a baseline and at least one postbaseline HDRS item 3 score. Endpoint analyses included all patients with at least a baseline score. Differences in mean values were compared between treatments using a conventional t-test. A pooled variance was used when the test of homogeneity of variance was not re- jected statistically, otherwise separate variance esti- mates were used. Differences in gender distribution between treatments were tested using Pearson’s chi- square test 1 df). Differences in proportions across all CDF values between treatments were tested using the Mantel-Haenszel chi-square test (1 df) for linear asso- ciation. Differences in the proportion of subjects who improved (CDF < O), experienced no change (CDF = 0), or worsened (CDF > 0) between treatments were tested using Pearson’s chi-square test (2 df). The rela- tionship between change in HDRS item 3 and the HDRS 17-item total score was assessed using Pearson’s correlation coefficient. All tests were two-sided and were considered statistically significant if the z   value was 2.0s. Analyses were conducted using SPSS. z RESULTS Data from a total of 8 clinical trials involving 243 patients were analyzed. Five trials involving 130 pa- tients compared a novel antidepressant vs. placebo (Pool 1) and four trials involving 123 patients assessed a novel antidepressant vs. a conventional TCA (Pool 2). Ten dothiepin-treated patients were included in both pools. Tables IA and IB display the composition and randomized patient numbers for each trial. Within Pool 1, study completion rates were similar between  76 zyxwvus ollefson zyxwvu t al. TABLE zyxwvutsr A. Composition of Pool 1 (novel antidepressants vs. placebo) zyxwvutsr Name of novel Novel antidepressant antidemessant (N7 Etoperidone Etoperidoneb Fluparoxan Nefazodone Dothiepin Total 9 11 11 40 10 81 zyxwvutsrqp Protocol EC. bProtocol EGO. TABLE 2A. Demographic and clinical attributes for Pool 1 Placebo N) 9 9 12 10 9 49 zyxwvut ~ ~ Novel Placebo N = 49) N = 81) Age (years) 49.3 t 13.7 45.8 * 14.3 Gender Number of males 37 19 Number of females 44 30 HDRS 17-item total 23.3 i 3.1 22.8 i 3.1 HDRS item 3 1.2 i 0.9 1.0 * 0.7 TABLE 1B. Composition of Pool 2 (novel antidepressants vs. TCAs) Novel Name of novel antidepressant TCA Etoperidoneb Imipramine 17 6 antidepressant Name of TCA N) zyxwv \r Adinazolam Desipramine 20 20 Dothiepin Doxepin 10 10 Fluoxetine Imipramine 19 21 Total 66 57 N = number ofpatients in the trial. bProtocol ETI. patients 71.6 novel vs. 79.6% placebo; x2 = 1.03, df = I, P = .311). Within Pool 2, statistically significantly more novel antidepressant-treated patients completed the entire protocol than their TCA counterparts (78.8 Tables IIA and IIB display demographic and clinical attributes by treatment group for Pools 1 and 2, re- spectively. There were no statistically significant treat- ment differences in gender, baseline HDRS 17-item total, or HDRS item 3 distribution in either pool or in age in Pool 1. There was a statistically significant treatment difference in age between patients random- ized to the novel antidepressants relative to the TCAs in Pool 2 (mean age 44.5 vs. 40.4), but this was not felt to be a clinically meaningful difference. Tables IILA and IIIB display the peak change differ- ence factors (CDF) (maximum HDRS item 3 score during the double-blind portion of the trial minus the baseline score; by treatment group for Pools 1 and 2, respectively). In Pool 1, there was a statistically signifi- cant treatment difference in mean peak CDFs (t = 2.17, df = 1.24, P = .032) favoring active therapy. Pa- tients treated with novel antidepressants demonstrated a lower mean peak CDF as a group (-.13 * .93) than those patients who received a placebo (.2l * .68). The analysis of percentage of patients across each peak CDF -2, -1, 0, 1, and 2) revealed a statistically sig- nificant difference between novel antidepressants and placebo (x = 4.57, df = 1, P = .032). The analysis of percentages across CDFs that indicate improvement, VS. 57.9%; x = 6.25, df = 1, P= 012). aValues represent the mean S.D. TABLE 2B. Demographic and clinical attributes for Pool 2 ~ ~ ~ Novel TCA N = 66 (N = 57) Age (years)* 44.5 t 11.7 40.4 * 10.5 Gender Number of males 32 25 Number of females 34 32 HDRS 17-item total 21.7 i 2.9 22.0 * 3.6 0.9 0.8 DRS item 3 0.8 0.8 'Values represent mean * S.D. 'Statistically significant treatment gvoup difference (P = .043 . no change, and worsening (<0, 0 ,and >O) also indi- cated a statistically significant difference (x = 6.46, df = 2, P = .039). As can be seen in Table IIIA, a higher percentage of patients treated with an antidepressant had peak negative CDFs which indicated improve- ment. Conversely, more placebo subjects had a posi- tive CDF, suggesting a greater chance of item 3 worsening from baseline sometime during the trial. In Pool 2, there were no statistically significant differ- ences in mean peak CDFs, in the percentages of patients achieving each peak CDF, nor in the percent- ages of patients in the improvement, no change, and worsening categories. Both the novel antidepressants and the conventional TCAs were associated with mean decreases in the peak HDRS item 3 score. The above analyses are focused on group and sub- group changes. In order to assess the emergence or in- tensification of substantial suicidal ideatiodacts, a search for change from a baseline HDRS item 3 score of 0 or 1 to 3 or 4 at any point in the study was con- ducted for each patient. No patient in any of the 8 tri- als experienced this definition of worsening at any point during the trials nor subjectively reported a clinical picture consistent with substantial worsening as an adverse event. Lastly, changes in HDRS item 3 and HDRS 17- item total from baseline to last measurement (end- point) were analyzed. In Pool 1, there was no statistically significant item 3 difference; however, the active agents (n = 81) were associated with a higher numeric  Research Article: Antidepressant-Related Change in Saicidality 77 TABLE 3A. zyxwvutsr hange in HDRS item 3 suicidal ideation from baseline t maximum value during treatment (pool 1)“ zyxwvutsrqpo Novel Placebo CDF (N= 78) (N=48) Suicidal ideation behavior score n zyxwvutsr  ) n ( ) Improved -2 5 (6.4) 0 -1 20 (25.6) 6 (12.5) No change 0 37 (47.4) 27 (56.3) Worsened 1 12 (15.4) 14 (29.2) Mean * S.D.* -.13 * .93 .21 * .68 zyxwvu   4 (5.1) l(2.1) zyxwvuts “CDF = change difference actor (HDRS item 3 maximum treatment score minus baseline score). *Statistically ignificant treatment difference (P = ,032). TABLE 3B. Change in suicidal ideation from baseline to maximum value during treatment (Fool 2)a ~ Novel TCA CDF (N=62) (N=53) Suicidal ideation behavior score n ( ) n ( ) Improved -3 l(1.6.) 0 -2 3 4.8) 3 5.7) -1 14(22.6) 9 (17.0) No change 0 36 (58.1) 33 (62.3) Worsened 1 7 (11.3) 8 (15.1) Mean t S.D.* -0.23 * 0.84 -0.13 * 0.74 2 l(1.6) 0 “CDF = change difference actor (HDRS item 3 maximum treatment score minus baseline score). score reduction than placebo (n = 49) (-.63 vs. -.39; t = 1.41, df = 128, P = .160). There was a trend for greater improvement in the mean HDRS 17-item total scores for those subjects randomized to an antidepressant rather than to placebo (-11.7 vs. -9.3; t = 1.87, df = 128, P = .064). In Pool 2, there was no statistically sig- nificant difference between the novel antidepressants (n = 66) and conventional TCAs (n = 57) on item 3 zyxwvu -.SS vs. -.46; t = -.57, df = 121, P = 369) or total score (-9.8 vs. -8.9; t = -.70, df = 121, P .484) changes from baseline to endpoint. Correlations of baseline to last measurement change in HDRS 17-item total score and HDRS item 3 were conducted. Of note, in both treatment groups, in each pool, change in HDRS item 3 was positively corre- lated with a change in HDRS 17-item total score sug- gesting item 3 variation was related to overall change in depression severity (correlation coefficients ranged from .39 to .58; P < .O1 in all four cases). Correla- tion coefficients were also calculated after the item 3 score was subtracted from the 17-item total score. The correlation coefficients remained positive and statistically significant (ranged from .27 to .49; P < .03 in all four cases). DISCUSSION MD is a hiphlv Drevalent disorder with a natural il history inclusive of potential morbidity/mortality.26 Prospective, double-blind, placebo-controlled investi- gations of suicidal ideation during the pharmacologic treatment of ambulatory depressives have been lim- ited. Montgomery and Pinder,” in a 1987 review, found “no acceptable evidence” for increased suicidal- ity attributable to any particular antidepressant. In re- minding us that “suicidal thoughts are an integral part of depression,” they emphasized that double-blind trials are necessary to demonstrate causality. More recently Mann and KapurZ8 eviewed the topic of suicidality during antidepressant pharmacotherapy. Their review spanned a wide variety of antidepressant compounds and failed to implicate any one selective neurotransmitter. In a study of relative risk, Fava and Rosenbaum3 sur- veyed 2 7 psychiatrists regarding 1 O 17 depressed pa- tients under their treatment (19% were suicidal at the initiation of treatment). This retrospective survey in- dicated that suicidal ideation did emerge during pharmacotherapy in approximately 1% of the de- pressed patients. However, no significant difference in incidence rates were reported between solo use of a selective 5-HT reuptake blocker (Fluoxetine 3 So/, , mixed amine tricyclic antidepressants (TCA) alone or with lithium (1.3 ), or among several other miscella- neous agents (3.0%). These data are concordant with another recent review which found no compelling data to suggest that any one antidepressant class carried a greater inherent risk of treatment-associated suicidality than any other.29 To date, large scale, double-blind clinical trials have shown that placebo recipients mani- fest similar HDRS item 3 worsenings as their active therapy counterpart^.^ -^^ To further explore treatment-associated changes in suicidality between types of antidepressants, we con- ducted this current retrospective analysis on a series of related double-blind/placebo or comparator con- trolled trials among ambulatory unipolar MD patients at a single investigative site. The benefit of a placebo- controlled sample Pool 1) was the provision of a “base rate” of HDRS item 3 changes during the comparable time interval. One limitation in this study, which is common to several other large scale, placebo-con- trolled trials was the exclusion of currently active, high risk suicide cases. Admittedly, their outcomes during pharmacotherapy may have been different. Clearly the ethical issue in such a patient receiving placebo or an investigational agent support their exclusion. Re- gardless, the target population studied here permit- ted testing of the primary study question; how do antidepressant vs. placebo-treated subjects compare relative to their rate of emergent suicidal ideation from baseline during treatment? This trial relied on the third item from the HDRS to assess change in suicidality. The HDRS represents a
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks