Maintenance Phase Efficacy of Sertraline for Chronic Depression

Transcription

1 Original Contribution Maintenance Phase Efficacy of for Chronic Depression A Randomized Controlled Trial Martin B. Keller, MD; James H. Kocsis, MD; Michael E. Thase, MD; Alan J. Gelenberg, MD; A. John Rush, MD; Lorrin Koran, MD; Alan Schatzberg, MD; James Russell, MD; Robert Hirschfeld, MD; Daniel Klein, PhD; James P. McCullough, PhD; Jan A. Fawcett, MD; Susan Kornstein, MD; Lisa LaVange, PhD; Wilma Harrison, MD; for the Chronic Depression Study Group Context. The chronic form of major depression is associated with a high rate of prevalence and disability, but no controlled research has examined the impact of long-term treatment on the course and burden of illness. Objective. To determine if maintenance therapy with sertraline hydrochloride can effectively prevent recurrence of depression in the high-risk group of patients experiencing chronic major depression or major depression with antecedent dysthymic disorder ( double depression ). Design. A 76-week randomized, double-blind, parallel-group study, conducted from September 1993 to November Setting. Outpatient psychiatric clinics at 10 academic medical centers and 2 clinical research centers. Intervention. Maintenance treatment with either sertraline hydrochloride in flexible doses up to 200 mg or placebo (n = 84). Patients. A total of 161 outpatients with chronic major or double depression who responded to sertraline in a 12-week, double-blind, acute-phase treatment trial and continued to have a satisfactory therapeutic response during a subsequent 4- month continuation phase. Main Outcome Measure. Time to recurrence of major depression. Results. afforded significantly greater prophylaxis against recurrence than did placebo (5 [6%] of 77 in the sertraline group vs 19 [23%] of 84 in the placebo group; P =.002 for the log-rank test of time-to-recurrence distributions). Clinically significant depressive symptoms reemerged in 20 (26%) of 77 patients treated with sertraline vs 42 (50%) of 84 patients who received placebo (P =.001). With use of a Cox proportional hazards model, patients receiving placebo were 4.07 times more likely (95% CI, ; P =.005) to experience a depression recurrence, after adjustment for study site, type of depression, and randomization strata. Conclusions. Maintenance therapy with sertraline is well tolerated and has significant efficacy in preventing recurrence or reemergence of depression in chronically depressed patients. JAMA. 1998;280: From the Departments of Psychiatry, Butler Hospital, Brown University, Providence, RI (Dr Keller); Cornell University School of Medicine, New York, NY (Dr Kocsis); University of Pittsburgh and Western Psychiatric Institute, Pittsburgh, Pa (Dr Thase); University of Arizona, Tucson (Dr Gelenberg); University of Texas Southwestern Medical Center, Dallas (Dr Rush); Stanford University School of Medicine, Stanford, Calif (Drs Koran and Schatzberg); University of Texas Medical Branch at Galveston (Drs Russell and Hirschfeld); State University of New York at Stony Brook (Dr Klein); Medical College of Virginia, Virginia Commonwealth University, Richmond (Drs McCullough and Kornstein); Rush Institute for Mental Well-Being, Chicago, Ill (Dr Fawcett); Quintiles, Research Triangle Park, NC (Dr LaVange); Pfizer Inc, and College of Physicians and Surgeons, Columbia University, New York, NY (Dr Harrison). Authors financial disclosures are listed at the end of this article. Reprints: Martin B. Keller, MD, Chair, Department of Psychology and Human Behavior, Brown University/ Butler Hospital, 345 Blackstone Blvd, Providence, RI AN ACCUMULATING body of research documents the enormous personal, psychosocial, and economic costs of depression. These costs result from reduced capacity to perform normal social roles such as working and parenting. 1-3 Depressed persons also experience diminished quality-of-life dimensions 4 and may use medical services excessively. 5 When compared with other medical diseases, the recent World Health Organization Global Burden of Disease study 6 identified depression as being 1 of the 4 most disabling illnesses in the world. Among the factors contributing to the profound level of disability associated with depression are high prevalence, increased rates of recurrence, chronicity, and associated medical and psychiatric comorbidity. 7 Epidemiologic surveys confirm that a significant proportion of those diagnosed as having depression (either major or minor) have a chronic course, with lifetime prevalenceestimatesthatareabove5%. 7,8 Early discontinuation of antidepressant therapy has been associated with a 25% relapse rate within 2 months, even among patients who do not have chronicity. 9 By 6 months, a 2- to 3-fold difference in relapse rates consistently favors the continuation of pharmacotherapy compared with placebo Thus, continuation of antidepressant therapy (defined as the 4-9 months of treatment immediately following acute treatment response) is now considered the standard of care to prevent relapse Optimal clinical management of depressive illness is not established once 6 to 9 months of continuation therapy are completed. A history of prior depressive recurrences, subsyndromal symptoms, JAMA, November 18, 1998 Vol 280, No. 19 Efficacy of for Chronic Depression Keller et al 1665

2 depression secondary to other comorbid psychiatric disorders, socioeconomic status, severity of depression, and chronicity of depression are all risk factors that predict a higher likelihood of recurrent depression The presence of 1 of these risk factors suggests that further maintenanceantidepressanttherapymightbe required. Several studies provide strong support for the prophylactic efficacy of maintenance antidepressant therapy in patients with a history of multiple prior depressive episodes. 11,16-18 But no adequately designed studies are available that examine the efficacy of maintenance therapy in the common subgroup of patients with chronic depression of at least 2 years duration. Currently, the majority of all new prescriptions written for depressed patients are for the selective serotonin reuptake inhibiting (SSRI) class of antidepressants. 19 Our study was designed to provide the first randomized placebo-controlled test of the safety and sustained prophylactic efficacy of maintenance treatment with an SSRI antidepressant, in this case sertraline hydrochloride, in the high-risk patient group who have been diagnosed as having chronic depression. METHODS Patient Population Subjects were patients who had responded to sertraline after participation in 1 of 2 prospective, 12-week acute treatment studies. The results of the 12- week study are presented in a separate report. 20 The maintenance study reported here was conducted at 10 academic centers and 2 clinical research centers from September 1993 to November Outpatients meeting a structured clinical interview diagnosis of chronic major depression (of at least 2 years duration), or dysthymic disorder with a concurrent diagnosis of major depression ( double depression ) based on the Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition (DSM-III-R) were enrolled. Patients who had a minimum baseline severity of 18 on the 24-item Hamilton Depression scale (HAM-D) after a 1- week single-blind placebo lead-in were initially entered into the 3-phase protocol. During the acute phase, patients were randomized in the double-blind study to 12 weeks of treatment with either sertraline hydrochloride or imipramine hydrochloride in a 2:1 ratio. Patients next entered a 4-month continuation phase if, at the end of the acute phase, they fell into either (1) the full remission response category, which is operationally defined as achieving a final HAM-D total score of 7 or less and a Clinical Global Impressions (CGI) improvement score of 1 or 2 (very much or much improved) or (2) the satisfactory therapeutic response category, which is operationally defined as achieving a reduction of 50% in the final HAM-D total score, as well as a score of 15 or less, a CGI improvement score of 2 or less, and a CGI severity score of 3 or less. Of the patients who received the sertraline treatment, 209 (49.1%) of 426 both completed 12 weeks of acute treatment and met full or satisfactory response criteria. These 209 patients were eligible for entry into the continuation phase of treatment. Patients were eligible to enter the maintenance phase reported if they had sustained at least a satisfactory antidepressant response to sertraline through the end of continuation therapy. Thesubjectsenteringthemaintenance phase represent only the subset of subjects who had been treated with sertraline continuously since they entered into the acute phase of the study. For patients who received imipramine during the acute treatment phase, 98 (46.9%) of 209 both completed the 12 weeks of acute treatment and met the full or satisfactory response criteria. To maintain blinding, this group of patients continued (as a parallel but nonrandomized group) receiving imipramine during subsequent continuation and maintenance phases. Those taking either sertraline or imipramine who failed to improve during the acute phase treatment were entered into the crossover phase of the study, with sertaline nonresponders switched to imipramine, and imipramine nonresponders switched to sertraline. The integrity of the study s double-blind component was not compromised. Thirty(60%) of51patientstreated with sertraline in the crossover phase of the study responded, while 52 (44%) of 117 patients treated with imipramine during the crossover phase responded. These crossover patients have also not been included in this report because the differential sequencing of study treatments made them noncomparable. Results from these parallel continuation and crossover phases of treatment will be the subject of separate reports. Study Design The maintenance study was approved by the institutional review board at each of the 12 collaborating centers. Separate written informed consent was obtained for the maintenance phase. The design of this phase consisted of random, doubleblind assignment to parallel groups for 76 weeks of treatment with either sertraline hydrochloride (in a flexible daily dose of 50 to 200 mg) or placebo. Patients randomized to the latter group tapered sertraline use via placebo substitution at a 50 mg/wk rate of reduction. Randomization was stratified by high and low probability of recurrence based on the 2 following variables hypothesized to be predictors of increased probability of depression recurrence: (1) presence of residual depressive symptoms (operationally defined as a HAM-D score 10 and a CGI severity score 3) at the end of continuation treatment; and (2) a history of 2 or more prior major depressive episodes. Outcome Measures Patients were evaluated and rated every 2 weeks for the first 12 weeks, and then monthly thereafter. If patients depression worsened, their visit frequency could be increased to once a week. Assessment instruments included clinician-rated instruments designed to systematically measure various aspects of the antidepressant s efficacy, which included the 24-item version of the HAM-D, 21 which assesses core, as well as associated or secondary, symptoms of depression; the Montgomery-Åsberg Depression Rating Scale, 22 which assesses core symptoms of depression; the Cornell Dysthymia Scale, 23 which assesses dysthymic symptoms; and the CGI severity and improvement scales. 24 These measures, together with the patient-rated Beck Depression Inventory, 25 which assesses severity of depression with an emphasis on cognitive depressive symptoms, and the Patient Global Evaluation 24 were completed on a monthly basis. In addition, the HAM-D and CGI scales were completed every 2 weeks during the first 12 weeks of the study. Various scales assessing quality of life and psychosocial functioning were also administered. Details on these measures and the effect of study treatment on these measures will be presented in a later article. Definition and Management of Exacerbations and Recurrences Reports of depression exacerbations triggered an appointment within 1 week. Patients were considered to have an impending recurrence of depression if, at either a scheduled or unscheduled assessment visit, they met all the following criteria: (1) DSM-III-R criteria for major depression for at least 3 weeks; (2) CGI severity score of 4 or more (at least moderate severity); (3) CGI improvement score of 3 or more (minimally improved or less); and (4) an increase in HAM-D to a score of 4 or more points higher than the maintenance phase baseline. Such patients were rescheduled for a second visit within 1 week (total dura JAMA, November 18, 1998 Vol 280, No. 19 Efficacy of for Chronic Depression Keller et al

3 tion of clinical worsening criteria of at least 4 weeks) and were declared to have experienced a recurrence if they continued to meet all the above criteria and if a senior investigator who interviewed the patient had concurred with the conclusion that the patient had experienced a recurrence of major depression. Patients meeting recurrence criteria could continue in the study if both patient and study physician agreed that no change in the study medication was indicated at that time. Instead, an increase in daily dose was undertaken at a rate of 50 mg/wk up to the maximum daily dose of 200 mg of sertraline hydrochloride. A similar double-blind titration was also used for patients receiving placebo treatment. Patients experiencing exacerbations of depression but who did not meet strict recurrence criteria also were managed using this procedure. If this titration procedure failed or if the patient was already receiving the maximum tolerable or permissible dose and a change of treatment was felt to be indicated, then patients who met recurrence criteria were discontinued from the study. Definition of Primary and Secondary End Points Three time-to-event outcome variables constituted the primary focus of the research reported herein. The first variable, specified as the primary end point of the trial a priori was time to recurrence of a major depressive episode. We also used 2 less stringent, post hoc end points because they reflected clinically relevant trigger points for treatment interventions in the long-term management of depression. They are time to reemergence of clinically significant depression and time to reemergence of first symptoms of depression. Establishment of the 2 less stringent end points was accomplished after a blinded review of HAM-D, GCI, and an overall clinical picture of all patients who discontinued the study prematurely. Agreement among 6 (75%) of 8 senior investigators was required for a patient to be categorized as having met the 2 post hoc clinical end points. Statistical Methods Baseline homogeneity of the 2 intentto-treat randomization groups was confirmed with statistical tests appropriate for continuous and categorical variables, always taking into account the effect of study site and randomization stratification factor. Those who responded to sertraline and who were entering the maintenance phase (both randomization groups combined) were also compared with patientsoriginallyrandomizedtoreceivesertraline at the start of the acute phase but whowerenotenteringmaintenancephase, with respect to the same background, clinical, and demographic variables. We applied the analysis of variance (ANOVA) and Cochran-Mantel-Haenszel methods to these comparisons, adjusting for both the study site and for the stratification factor. The primary objective of the maintenance phase was to compare the efficacy and safety of continuing sertraline treatment vs placebo substitution in preventing recurrence of major depression in patients experiencing chronic depression. The primary efficacy end point, defined a priori, was time to recurrence of major depression. Secondary efficacy end points, also defined a priori, included change from baseline in the symptom severity ratings. These assessments were obtained at multiple visits throughout the maintenance phase. The results of these scores are reported herein for the patient s last or end point assessment for each severity rating. Secondary end points also included the post hoc, time-to-event variables, consisting of time to reemergence and time to first signs of depression. These were post hoc end points that were defined (prior to breaking the study blind) to be confirmatory of the primary outcome as a consequence of the low rates of protocol-defined recurrence overall in the study. These secondary, post hoc end points were established as a more sensitive measure of outcome that could detect a treatment difference in the absence of the more stringent primary end points. The primary analysis consisted of a comparison of the time to recurrence of depression for patients taking sertraline and patients taking placebo based on the log-rank 2 test statistic for differences in time-to-failure distributions, which were stratified on the probability of recurrence. Patients who discontinued the study early were censored at their time of discontinuation. Kaplan- Meier method estimates of the time-tofailure distributions were produced for descriptive purposes. Median failure times and estimates of the cumulative failure probabilities at 8, 24, 52, and 76 weeks of maintenance therapy were derived from the Kaplan-Meier distributions, also for descriptive purposes. The secondary time-to-failure outcomes (time to reemergence and time to first signs of depression) were similarly analyzed. Secondary efficacy measures corresponding to the specified symptom severity ratings were analyzed for between-treatment-group differences at each patient s last, or end point visit. Comparisons were made in the context of analysis of covariance (ANCOVA) models fit to change-from-baseline values, which were adjusted for study site, randomization, stratification factor, and maintenance phase baseline values. Siteby-treatment interactions were assessed in a second model fit to each outcome. The changes within treatment groups from the maintenance phase baseline to the end point with respect to each outcome measure were assessed by 1-sample paired t tests. Secondary analyses of time to recurrence and time to reemergence of clinically significant depression were conducted to assess the effects of additional covariatesontreatmentfailure. Adjusted relative risk ratios for treatment were calculatedbasedoncoxproportionalhazards regression models that included effects for pooled study site, type of depression(chronicmajorvsdoubledepression), and randomization strata (high vs low risk of recurrence). Additionally, proportional hazards regression models were fit in a forward stepwise fashion to assess the importance of a wide range of potential predictors on time to recurrence and time to reemergence. Potential interactions of significant predictors with treatment were assessed in the context of these reduced models. For all proportional hazards models used to estimate adjusted relative risks, the assumption of proportionality was confirmed via tests of interactions between time to outcome and treatment and between time to outcome and significant covariates. Results of an exploratory analysis, specified post hoc, examining the effects of withdrawal from sertraline treatment were based on the Fisher exact test for differences in the proportions of patients in the sertraline and placebo groups who experienced recurrence in the first 4 weeks following randomization. RESULTS Demographic and Clinical Characteristics Among the 209 patients who responded to sertraline and who began the 4-month continuation phase, 179 of them completed the continuation phase, 169 met criteria for remission or satisfactory response in the opinion of the investigator, and 161 enrolled in the maintenance treatment trial. These 161 patients constitute the analysis sample for all the data analyses presented in this study. They were assigned to receive either sertraline or placebo (n = 84). Figure 1 schematically summarizes the study design and the patients available at each phase of the study. At the end of the continuation trial, there were no significant differences in efficacy measurements between patients diagnosed as having chronic ma- JAMA, November 18, 1998 Vol 280, No. 19 Efficacy of for Chronic Depression Keller et al 1667

4 Randomized to Hydrochloride Acute Phase (N = 426) 76 wk of Maintenance Monthly Measures of Outcome Withdrawn (n=42) Intervention Ineffective (n = 11) Intercurrent Illness/Laboratory Abnormality (n = 0) Protocol Violation (n = 8) Lost to Follow-up: Other Reasons (n = 15) Adverse Events (n = 8) Completed Trial (n = 35) Figure 1. Study flow diagram. Nonresponders (Crossover) (n = 117) Acute Phase Dropouts (n = 76) Completers not Continuing Into the Continuation Phase (n = 24) Responders Who Continued Into the Continuation Phase (n = 209) Eligible Patients, Satisfactory Response at End of Continuation Phase (n = 169) (n = 84) 76 wk of Maintenance Monthly Measures of Outcome Completed Trial (n = 24) Continuation Phase Dropouts (n = 30) Completers not Continuing Into Maintenance Phase (n = 10) Elected not to Continue Into Maintenance Phase (n = 8) Withdrawn (n = 60) Intervention Ineffective (n = 34) Intercurrent Illness/Laboratory Abnormality (n = 2) Protocol Violation (n = 4) Lost to Follow-up: Other Reasons (n = 19) Adverse Events (n = 1) jor depression or double depression. Consequently, these patient groups were combined. Table 1 summarizes the demographic and clinical characteristics measured at baseline for patients randomized to receive sertraline treatment at the start of the acute phase but excluding patients entering the crossover treatment group. Summaries are shown both for patients entering the maintenance phase by treatment group (n = 77 for sertraline; n = 84 for placebo; N = 161, total), and for patients discontinuing prior to the maintenance phase (n = 148). There were isolated differences between these 2 groups, including the tendency of patients entering the maintenance phase to be better educated (P =.03), more likely to be married (P =.01), and to have had prior psychotherapy (P =.03) than patients not entering the maintenance phase. The 2 maintenance phase treatment groups were also similar in the variables we examined, with a few exceptions. Patients in the placebo group were less likely to have a past history of panic disorder (P =.02) and, among the patients with double depression, placebo patients reported a shorter duration of current major depression than those randomized to the sertraline group (P =.03). Study Treatment Effects of Discontinuation After 7 Months of Acute Plus Continuation-phase Treatment. The effectoftapereddiscontinuationofsertraline during the first 4 weeks of maintenance phase treatment was examined to assess whether medication withdrawal accelerated the recurrence rates. Eleven patients (13%) in the placebo group experienced a reemergence of depression during the first 4 weeks after randomization compared with 7 (9%) of those who continued taking sertraline (P =.55). The mean final dose of sertraline hydrochloride was mg. The mean final number of placebo pills taken was 3.4. The mean final sertraline hydrochloride dose in the maintenance phase did not differ significantly from the mean final dose in the acute treatment phase (141.6 mg) (n = 77; P =.47 from 1-sample t test; not significant). Adverse Events and Attrition. In general, maintenance treatment with both sertraline and placebo were well tolerated, lending support to the integrity of the double-blind. There were no differences overall in the incidence of adverse event reporting between the 2 treatment groups (62/84 patients taking placebo vs 62/77 patients taking sertraline reported at least 1 adverse event; P =.31). Table 2 presents treatment-emergent adverse eventsreportedinatleast10% ofpatients ineithertreatmentgroup. Significantdifferences were observed with respect to pain and sexual dysfunction, with more patients treated with sertraline reporting sexual dysfunction and more patients treatedwithplaceboreportingpain. Only 8 patients (10.4%) treated with sertraline, compared with 1 patient (1.2%) treated with placebo, dropped out citing adverse events as the reason (P =.01; Fisher exact test). Other reasons cited for dropping out of the study that were unrelated to recurrence or depression symptom reemergence were similar for both groups. Only attrition due to insufficient response and attrition due to adverse events were statistically significant. Attrition related to consensus determination of treatment failure is reported below. Treatment Efficacy End Point Analysis of Symptom Severity Ratings for the Intent-to-Treat Sample. Table 3 provides mean scores at entry into the acute and maintenance phases and at end point. No significant differences were detected between treatment groups at entry into the maintenance phase. By contrast, maintenance sertraline resulted in significantly better outcomes on all 6 symptom measures. This reflects a broad improvement across all symptom dimensions of depressive illness, including cognitive (Beck DepressionInventory), core(ham-d, Montgomery-Åsburg Depression Rating Scale), and accessory (HAM-D) symptoms of major depression, symptoms of dysthymia (Cornell Dysthymia Scale), and global symptoms (CGI). There were no significant treatment-by-center interactions. Treatment Efficacy: Recurrence Rates and Time to Recurrence of Major Depression. Table 4 summarizes major depression recurrence rates observed during maintenance treatment by the a priori definition of recurrence and consensus ratings. The stringent definition (requiring 3 weeks of treat JAMA, November 18, 1998 Vol 280, No. 19 Efficacy of for Chronic Depression Keller et al

5 Table 1. Demographic and Clinical Data on Study Patients at Entry Into the Initial Acute Phase of Treatment Compared With Patients Treated With Hydrochloride in the Acute Phase Who Did Not Continue Into the Maintenance Phase* Prescription Maintenance Phase Patients Not Reaching Total Patients at Patient Variable (n = 84) Maintenance (n = 148) Acute Phase (N = 309) Female, No. (%) 48 (62) 58 (69) 86 (58) 192 (62) Age, mean (SD), y 40.8 (9.0) 42.4 (9.7) 39.7 (10.7) 40.7 (10.1) Married, No. (%) 36 (47) 37 (44) 43 (29) 116 (38) Education, No. (%) At least high school graduate 72 (95) 83 (99) 141 (96) 296 (96) At least college graduate 29 (38) 35 (42) 41 (28) 105 (34) Depression diagnosis Chronic major, No. (%) 40 (52) 36 (43) 69 (47) 145 (47) Duration of current episode, mean (SD), y 9.9 (11.6) 8.0 (8.7) 8.7 (7.8) 8.9 (9.2) Double depression, No. (%) 37 (48) 48 (57) 79 (53) 164 (53) Duration of current major depressive disorder, mean (SD), y 4.6 (7.5) 2.0 (2.8) 2.9 (6.4) 3.1 (5.9) Duration of current dysthymia, mean (SD), y 23.9 (12.5) 24.7 (13.2) 21.9 (13.6) 23.2 (13.2) Age at onset of depression, mean (SD), y 24.9 (11.2) 25.7 (12.5) 24.5 (11.0) 24.9 (11.5) No. of prior episodes of depression, mean (SD) 1.8 (2.2) 1.9 (2.1) 1.8 (2.0) 1.8 (2.1) Patients with 2 previous episodes, No. (%) 37 (48) 39 (46) 63 (43) 139 (45) History of comorbid anxiety disorder, No. (%) Panic disorder 12 (16) 4 (5) 7 (5) 23 (8) Social phobia 8 (10) 11 (13) 21 (14) 40 (13) Generalized anxiety disorder 2 (3) 2 (2) 9 (6) 13 (4) Any anxiety disorder 23 (30) 18 (22) 38 (26) 79 (26) History of alcohol abuse, No. (%) 28 (36) 20 (24) 42 (29) 90 (29) History of substance abuse, No. (%) 33 (43) 25 (30) 52 (35) 110 (36) Comorbid Axis II personality disorder, No. (%) 34 (44) 42 (51) 69 (47) 145 (47) Prior treatment with antidepressants Adequate treatment, No. (%) 19 (25) 15 (18) 27 (18) 61 (20) Prior psychotherapy, No.(%) 44 (57) 57 (68) 73 (49) 174 (56) *See Figure 1 for the disposition of patients during the course of study treatment (426 initial sertraline patients minus 117 crossover patients equals the 309 total sertraline patients displayed in the fourth data column). Numbers exclude crossover patients. Diagnosed by Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition (DSM-III-R) criteria. Adequate treatment defined as 150 mg or more of imipramine hydrochloride or its equivalent for at least 1 month. ment and a confirmatory second evaluation 1 week later by an independent senior investigator) yielded a low rate of protocol-defined recurrences at end point for both sertraline (6% [5/77]) and placebo(23%[19/84]). This unexpectedly low rate of recurrence was probably due to attrition from the study prior to meeting stringent recurrence criteria. The time to recurrence was significantly delayed for patients treated with sertraline compared with those treated with placebo (P =.002), stratifying for patients who are at high risk vs those who are at a low risk of recurrence. Similar significant benefit for sertraline was found using the less stringent consensus definitions of reemergence of depression symptoms (P =. 001) and first symptoms of depression (P<.001). Only 20 (26%) of the 77 patients taking sertraline experienced reemergence of depression compared with 42 (50%) of 84 patients taking placebo, and only 26 (34%) of 77 patients taking sertraline showed first signs of Table 2. Most Common Treatment-Emergent Adverse Events During the Maintenance Phase: Incidence of at Least 10% in Either the Hydrochloride or Treatment Groups* Coded Term for Adverse Event No. (%) (n = 84) Headache 22 (28.6) 27 (32.1) Insomnia 15 (19.5) 11 (13.1) Sexual dysfunction 13 (16.9) 2 (2.4) Sweating increased 12 (15.6) 6 (7.1) Weight increase 12 (15.6) 6 (7.1) Diarrhea 12 (15.6) 6 (7.1) Mouth dry 10 (13.0) 7 (8.3) Nausea 10 (13.0) 5 (6.0) Dizziness 9 (11.7) 13 (15.5) Fatigue 9 (11.7) 13 (15.5) Dyspepsia 9 (11.7) 10 (11.9) Pain 1 (1.3) 9 (10.7) *Includes only treatment-related adverse events that started in the maintenance phase (excludes other known causes and concurrent illness ). Includes ejaculatory delay (0% sertraline, 1.2% placebo), anorgasmia (3.9% sertraline, 0% placebo), impotence (5.2% sertraline, 0% placebo), or loss of libido (9.1% sertraline, 1.2% placebo). Significant at P.01. Significant at P.05. depression compared with 50 (60%) of 84 patients taking placebo. Table 5 presents the cumulative survival probabilities for recurrence and reemergence of depression at periodic intervals during maintenance follow-up. By the eighth week of maintenance treatment, the survival distributions began to separate, with sertraline providing patients with significantly better prophylaxis than placebo at all points thereafter. Themediantimetorecurrenceorreemergenceisnotdefinedforsertralinepatients because less than 50% of these patients experienced either event. The median survival time to reemergence among patients taking placebo was 26 weeks (95% confidence interval [CI], 14 to undefined). Figure 2 shows the Kaplan-Meier estimates of the survival distributions for both treatment groups, with the 3 panels illustrating comparisons for the primary and secondary time-to-failure outcomes. The adjusted relative risk of placebo treatment was estimated under the assumption of proportional hazard rates for time to recurrence and time to reemergence of depression. Patients receiving placebo were 2.18 (95% CI, ) times as likely to experience reemergence of depression, and 4.07 (95% CI, ) times as likely to experience depression recurrence as patients taking sertraline during maintenance therapy, adjusting for pooled study site, type of depression, and randomization strata (P =.005 for both outcomes). Attrition also was significantly higher among patients receiving placebo than among those taking sertraline. Approximately 6 months into the maintenance phase, almost half of the patients taking placebo had discontinued the study, compared with only 31% of patients taking sertraline. By study end, 24 (29%) of 84 patients taking placebo remained in the study, compared with 35 (45%) of 77 patients taking sertraline. The median time of discontinuation among patients taking placebo was 24 weeks (95% CI, 15-40) compared with 53 weeks (95% CI, 32 to undefined) among patients taking sertraline (P.02). Effects of Clinical Covariates on Symptomatic Reemergence and Episode Recurrence. We examined the potential influence of a number of clinical variables on recurrence and reemergence rates by fitting Cox proportional hazards regression models to time-to-event data. Various clinical covariates were entered into a stepwise proportional hazards regression model to determine the following: (1) whether any of the covariates considered were significant predictors of time to reemergence and time to recurrence and (2) whether any of the covariates considered had a significant JAMA, November 18, 1998 Vol 280, No. 19 Efficacy of for Chronic Depression Keller et al 1669

6 Table 3. Acute Phase Baseline, Maintenance Phase Baseline, and Maintenance Phase End Point Scores for Efficacy Measures* Baseline Scores, Mean (SD) Hydrochloride (n = 84) Hamilton Depression Scale (24-item) Acute phase baseline 25.3 (5.0) 24.5 (4.1) Maintenance phase baseline 5.5 (4.2) 6.3 (3.7) Maintenance end point 9.9 (9.2) 15.3 (10.5) Change from maintenance baseline 4.3 (9.5) 8.9 (10.9).001 Clinical Global Impressions severity Acute phase baseline 4.2 (0.5) 4.0 (0.5) Maintenance phase baseline 1.5 (0.6) 1.4 (0.6) Maintenance end point 2.1 (1.3) 2.8 (1.5) Change from maintenance baseline 0.6 (1.3) 1.4 (1.5).001 Clinical Global Impressions improvement Acute phase baseline NA NA Maintenance phase baseline 1.3 (0.4) 1.2 (0.4) Maintenance end point 1.9 (1.3) 2.7 (1.5).001 Beck Depression Inventory Acute phase baseline 25.4 (8.7) 23.5 (8.0) Maintenance phase baseline 5.1 (4.8) 5.4 (4.8) Maintenance end point 9.0 (9.9) 13.9 (10.7) Change from maintenance baseline 3.7 (9.2) 8.5 (11.1).004 Cornell Dysthymia Acute phase baseline 41.5 (9.9) 39.6 (9.1) Maintenance phase baseline 8.5 (7.2) 10.2 (6.9) Maintenance end point 14.3 (14.9) 25.6 (18.4) Change from maintenance baseline 5.7 (14.4) 15.2 (18.9).001 Montgomery-Åsburg Depression Rating Scale Acute phase baseline 25.4 (7.2) 24.2 (7.1) Maintenance phase baseline 4.2 (4.2) 5.2 (4.4) Maintenance end point 7.9 (9.6) 15.0 (12.5) Change from maintenance baseline 3.7 (9.6) 9.7 (13.0).001 *No significant differences were found between treatment groups at acute or maintenance phase baseline, adjusting for pooled site and probability of recurrence, except with respect to Clinical Global Impressions severity at acute baseline (P =.03). NA indicates not applicable. P values comparing treatment groups for change from baseline are from an analysis of covariance (ANCOVA) adjusting for pooled site, probability of recurrence, and maintenance baseline values. P value for within-treatment group change from maintenance baseline from a 1-sample t test that is less than.05. Table 4. Rates of Major Depression During Maintenance Study Treatment* No. (%) Hydrochloride (n = 84) Log-Rank P Experienced recurrence by strict protocol criteria 5 (6) 19 (23).002 Experienced depression reemergence 20 (26) 42 (50).001 by consensus assessment Showed first symptoms of reemergence of depression by consensus assessment 26 (34) 50 (60).001 *Percentages are based on the number of patients randomized to each treatment. P values based on log-rank tests for comparisons of failure distributions for time-to-event data, stratifying on probability of recurrence. Table 5. Kaplan-Meier Estimates of Cumulative Time-to-Failure Probabilities for Patients Experiencing Depressive Illness Recurrence, and Depression Symptom Reemergence During Maintenance Phase Treatment* Weeks Treatment Recurrence of Depression (0-0.04) (0-0.10) (0-0.13) ( ) (n = 84) ( ) ( ) ( ) ( ) Reemergence of Depression Symptoms ( ) ( ) ( ) ( ) (n = 84) ( ) ( ) ( ) ( ) *Numbers in parentheses are 95% confidence intervals. P interaction with treatment. For reemergence of depression, longer duration of dysthymia was the only significant predictor (P =.004), although a trend was also evident for higher acute baseline CGI severity (P =.06). Controlling for duration of dysthymia, the treatment effect is still very strong (relative risk, 2.18; 95% CI, ; P =.005), indicating little or no confounding of the treatment effect from this covariate. Moreover, the 2-way interaction was not significant (P =.46). The only significant predictor of time to recurrence was higher acute baseline CGI severity (P =.01). Again, this effect did not diminish the treatment effect (relative risk, 5.13; 95% CI, ; P =.001), and the severity by treatment interaction was not significant (P =.99). The risk of reemergence of depression during placebo treatment among patients experiencing their first episode was relatively low (40% [10/25]) compared with patients with recurrent episodes of major depressive disorder (55% [32/58]). By contrast, sertraline was equally effective for patients in these subgroups with 9 (29%) of 31 first-episode patients experiencing a depression reemergence episode and 11 (24%) of 46 patients experiencing a depression reemergence. Although the drug-placebo difference in prophylactic efficacy was higher for patients with recurrent depression(ie, 31% vs 11%), the treatmentby-prior episode events interaction analysis was not statistically significant (P =.25) in predicting time to reemergence, more likely because of low power resulting from low event rates. Finally, it should be noted that entry criteria permitted patients with HAM-D scores as high as 15 to enter the maintenance phase treatment. In fact, only 24 patients (14.9%) had HAM-D total scores above 10 at the start of maintenance phase treatment, suggesting that only a small subset of patients entered with a substantial degree of residual depressive symptoms. Consistent with this, 80% of patients entered the maintenance phase with a sustained remission of at least 4 months duration (operationally defined by a Longitudinal Interval Follow-up Evaluation score 26 of 1 or 2 indicating at most mild residual depressive symptoms). Given this low level of residual depressive symptoms, it is not surprising that the presence of such residual symptoms was not found to predict a higher rate of depressive reemergence or recurrence. COMMENT This multicenter, placebo-controlled, randomized trial is the first to report on the maintenance phase efficacy of an 1670 JAMA, November 18, 1998 Vol 280, No. 19 Efficacy of for Chronic Depression Keller et al

7 Survival Distribution Function A B C Treatment Hydrochloride Time, wk Time, wk Time, wk Figure 2. A, time to recurrence of major depression. B, time to reemergence of symptoms of depression. C, time to first signs of depression reemergence. SSRI antidepressant in patients experiencing chronic depression. The results of this study demonstrate unequivocally that sertraline protects against recurrence or reemergence of depression and considerably extends the time in remission for this high-risk population. The results are consistent across all outcome measures and across 3 different criteria for recurrence or reemergence of depression. was very well tolerated during long-term use, although adverse events caused numerically higher and statistically significant attrition (n = 8) than did placebo (n = 1). One of the most widely accepted tenets of modern psychopharmacology is the value of maintenance treatment of major depression. And yet this de facto consensus treatment strategy remains largely an article of faith, resting as it does on a tenuous empirical base of a small number of placebo-controlled studies totaling fewer than 600 patients. In fact, the only published maintenance treatment studies of longer than 12 months used tricyclic or monoamine oxidase inhibitors ,27 The patients treated in our study not only had chronic depression but also had high levels of comorbidity, past substance abuse, moderate to high depressive symptom severity, and marked functional impairment. Most had been ill their entire adult lives. In light of such chronicity, severity, and comorbidity, relatively vigorous doses of sertraline hydrochloride (eg, an initial mean maintenance dose of 139 mg) were used. It should be noted that the current report summarizes results for the maintenance phase of a long-term treatment study. As a result, patients who were so sufficiently intolerant of sertraline that they discontinued treatment were not available for entry into this phase of the study. Similarly, patients were also not available who improved during the acute or continuation phases sufficiently enough that they were unwilling or unable to meet the commitments of an arduous maintenance protocol. Therefore, as favorable as the results are for sertraline in longterm antidepressant prophylaxis, it should be emphasized that the current results examine only a 38% subset (161/ 426) of the patients initially randomized to receive sertraline in the acute phase. Influence of Clinical Covariates on Outcome Severity of depression prior to the initial acute phase treatment was the only significant predictor of early recurrence of depression (P.02). Surprisingly, the number of prior episodes and the presence of residual depressive symptoms were not associated with an increased rate of depression recurrence, although patients with recurrent depression did have a higher recurrence risk while taking placebo, and sertraline did provide the greatest relative prophylactic benefit in this group. For reemergence of depression, longer duration of dysthymia was the only significant predictor (P =.004), although higher acute phase baseline depression severity was trending toward significance (P =.06). The influence of dysthymia as a covariate raises issues about the appropriateness of combining the patients diagnosed as having double depression and those with chronic major depression. However, no between-group differences existed at maintenance phase baseline and, in fact, the treatment effect observed for sertraline remained strong after controlling for dysthymia duration (P =.005). Was the Depression Recurrence Rate for Unusually Low? The reemergence of depression in patients treated with placebo was approximately 2-fold higher than what was observed for patients treated with sertraline. It is perhaps remarkable that nearly 50% of a patient sample entering the study with a mean of approximately 10 years of major depression, or 25 years of dysthymia, managed not to relapse during 18 months of placebo treatment. The repeated study visits and duration of contact with mental health professionals at each visit may have contributed to the relatively low recurrence rate in the placebo group, but it is possible that chronicity, if aggressively treated with acute and continuation therapy, is simply less strong of a risk factor for subsequent depression recurrence than is an established history of multiple prior recurrences. The recurrence risk associated with the latter is confirmed by maintenance treatment studies that have reported higher recurrence rates (70%- 80%) for patients receiving placebo, 11,16-18 as well as results from the National Institute of Mental Health Collaborative Depression Study, which identified a high number of prior episodes as a predictor of increased risk of recurrence. 16 Alternatively, it may be that the patients treated with an SSRI antidepressant such as sertraline are prone to discontinuation-emergent relapse, as has been proposed to occur on discontinuation of tertiary amine tricyclics. 28,29 The JAMA, November 18, 1998 Vol 280, No. 19 Efficacy of for Chronic Depression Keller et al 1671

8 results from the only published SSRI maintenance of treatment study 11 are consistent with this hypothesis, as they observed a 57% 12-month recurrence rate during placebo treatment despite an average of 3.8 prior episodes of depression. References 1. Mintz J, Mintz LI, Arruda MJ, Hwang SS. Treatments of depression and the functional capacity to work. Arch Gen Psychiatry. 1992;49: Klerman GL, Weissman MM. The course, morbidity, and costs of depression. Arch Gen Psychiatry. 1992;49: Greenberg PE, Stiglin LE, Finkelstein SN, et al. The economic burden of depression in J Clin Psychiatry. 1993;54: Wells KB, Stewart A, Hays RD, et al. The functioning and well-being of depressed patients: results from the Medical Outcomes Study. JAMA. 1989;262: Von Korff M, Ormel J, Katon W, Lin E. Disability and depression among high utilizers of health care: a longitudinal analysis. Arch Gen Psychiatry. 1992; 49: Murray CJ, Lopez AD. Global mortality, disability, and the contribution of risk factors: Global Burden of Disease Study. Lancet. 1997;349: Robins L, Regier DA, eds. Psychiatric Disorder in America. New York, NY: Free Press; Kessler RC, McGonagle KA, Zhao S, et al. Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: results from the National Comorbidity Survey. Arch Gen Psychiatry. 1994;51: Maj M, Veltro F, Pirozzi R, Lobrace S, Magliano L. Pattern of recurrence of illness after recovery from an episode of major depression: a prospective study. Am J Psychiatry. 1992;149: Prien RF, Kupfer DJ. Continuation drug therapy for major depressive episodes. Am J Psychiatry. 1986;143: Conclusion has previously been shown to be effective in the prevention of relapse in patients experiencing nonchronic episodes of major depression in a 44-week placebo-controlled continuation study. 12 Our study extends these results to an 18- month, placebo-controlled maintenancediscontinuation study that followed 7 months of acute and continuation treatment. The results of this study demonstrate the sustained efficacy of sertraline as a maintenance treatment for the prophylaxis of recurrence or reemergence of depression in patients who are at high risk for recurrences due to chronicity of depression, comorbidity, or established history of multiple previous episodes. Whether these results are specific to sertraline, or generalize to SSRI antidepressants as a class, will need to be clarified in further maintenance treatment studies. This study was supported by a grant from Pfizer Inc, New York, NY. Dr Keller is a consultant for Pfizer Inc; Bristol- Myers Squibb Co, Princeton, NJ; and Forest Laboratories, New York, NY. He has received grant research support from Wyeth-Ayerst Laboratories, Philadelphia, Pa; SmithKline Beecham Pharmaceuticals, Philadelphia; The Upjohn Co, Kalamazoo, Mich; Pfizer Inc; and Bristol-Myers Squibb Co. He serves on the advisory board for Wyeth-Ayerst Laboratories; Pfizer Inc; Bristol-Myers Squibb Co; Eli Lilly and Co, Indianapolis, Ind; Forest Laboratories; and Organon Inc, West Orange, NJ. He has received honoraria from Wyeth-Ayerst Laboratories, Pfizer Inc, Bristol-Myers Squibb Co, and Eli Lilly and Co. Dr Thase is a consultant for Bristol- Myers Squibb Co, Organon Inc, Pfizer Inc, and Wyeth-Ayerst Laboratories. He has received research grant support from Bristol-Myers Squibb Co; Eli Lilly and Co; Glaxo Wellcome Inc, Research Triangle Park, NC; Lipha Pharmaceutical, New York, NY; Organon Inc; Pfizer Inc; and Wyeth-Ayerst Laboratories. He is member of the speakers bureau for Bristol-Myers Squibb Co; Eli Lilly and Co; Glaxo Wellcome Inc; Pfizer Inc; SmithKline Beecham Pharmaceuticals; Solvay Pharmaceuticals Inc, Marietta, Ga; and Wyeth-Ayerst Laboratories. This study was completed with the cooperation and participation of the following investigators: Martin B. Keller, MD, Program Director, Gabor Keitner, MD, Ivan Miller, PhD, Brown University, Providence, RI; James H. Kocsis, MD, John C. Markowitz, MD, Cornell University, New York, NY; Daniel Klein, PhD, Fritz Henn, MD, David Schlager, MD, State University of New York, Stony Brook; James P. McCullough, PhD, Susan Kornstein, MD, Virginia Commonwealth University, Richmond; Robert M. A. Hirschfeld, MD, James Russell, MD, University of Texas, Galveston; Michael E. Thase, MD; Robert Howland, University of Pittsburgh, Pittsburgh, Pa; John S. Carman, MD, Psychiatry and Research Inc, Atlanta, Ga; A. John Rush, MD, George Trapp, MD, University of Texas Southwestern, Dallas; Alan F. Schatzberg, MD, Lorrin Koran, MD, Stanford University, Stanford, Calif; Alan Gelenberg, MD, Pedro Delgado, MD, University of Arizona, Phoenix; John Feighner, MD, Feighner Research Institute, San Diego, Calif; Jan A. Fawcett, MD, John Zajecka, MD, Rush Institute for Mental Well-Being, Chicago, Ill. Edward Schweizer, MD, New York, NY, provided editorial assistance. 11. Montgomery SA, Dufour H, Brion S, et al. The prophylactic efficacy of fluoxetine in unipolar depression. Br J Psychiatry. 1988;153(suppl 3): Doogan DP, Caillard V. in the prevention of depression. Br J Psychiatry. 1992;160: US Department of Health and Human Services. Depression Guideline Panel, Depression in Primary Care, Treatment of Major Depression. Vol 2. Washington, DC: Agency for Health Care Policy and Research; NIMH/NIH Consensus Development Conference Statement. Mood disorders. Am J Psychiatry. 1985;142: Coppen A, Mendelwicz J, Kielholz P. Pharmacotherapy of Depressive Disorders: A Consensus Statement. Geneva, Switzerland: World Health Organization; Prien RF, Kupfer DJ, Mansky PA, et al. Drug therapy in the prevention of recurrences in unipolar and bipolar affective disorders. Arch Gen Psychiatry. 1984;41: Frank E, Kupfer DJ, Perel JM, et al. Threeyear outcomes for maintenance therapies in recurrent depression. Arch Gen Psychiatry. 1990;47: Robinson DS, Lerfald SC, Bennett B, et al. Continuation and maintenance treatment of major depression with the monoamine oxidase inhibitor phenelzine. Psychopharmacol Bull. 1991;27: Prescription data for 1997, Ambler, Pa: IMS America. 20. Keller MB, Rush AJ, Thase ME, et al. Evaluating optimal treatment strategies in chronic depression. J Clin Psychiatry. 1998;59: Hamilton MA. A rating scale for depression. J Neurol Neurosurg Psychiatry. 1960;23: Montgomery SA, Åsberg M. A new depression scale designed to be sensitive to change. Br J Psychiatry. 1979;134: Mason BJ, Kocsis JH, Leon AC, et al. Assessment of symptoms and change in dysthymic disorder. In: Kocsis JH, Klein DN, eds. Diagnosis and Treatment of Chronic Depression. New York, NY: Guilford Press; 1995: Guy W. ECDEU Assessment Manual for Psychopharmacology. Washington, DC: National Institute of Mental Health,US Dept of Health, Education, and Welfare; 1976: Beck AT, Epstein N, Brown G, Steer RA. Psychometric properties of the Beck Depression Inventory: twenty-five years later. Clin Psychol Rev. 1988;8: Keller MB, Lavori PW, Friedman B, et al. The Longitudinal Interval Follow-up Evaluation: a comprehensive method for assessing outcome in prospective longitudinal studies. Arch Gen Psychiatry. 1987;44: Kupfer DJ, Frank E, Perel JM, et al. Five-year outcome for maintenance therapies in recurrent depression. Arch Gen Psychiatry. 1992;49: DilsaverSC.Withdrawalphenomenaassociated with antidepressant and antipsychotic agents. Drug Saf. 1994;10: Littrell J. Relationship between time since reuptake-blocker antidepressant discontinuation and relapse. Exp Clin Psychopharmacol. 1994;2: JAMA, November 18, 1998 Vol 280, No. 19 Efficacy of for Chronic Depression Keller et al