ORIGINAL ARTICLE. The Long-term Course of Rapid-Cycling Bipolar Disorder

Transcription

1 ORIGINAL ARTICLE The Long-term Course of Rapid-Cycling Bipolar Disorder William Coryell, MD; David Solomon, MD; Carolyn Turvey, PhD; Martin Keller, MD; Andrew C. Leon, PhD; Jean Endicott, PhD; Pamela Schettler, PhD; Lewis Judd, MD; Timothy Mueller, MD Background: Rapid cycling among patients with bipolar affective disorders is important because of its implications for long-term prognosis and for the use of antidepressants. To our knowledge, no prospective study has, as yet, described the course of this phenomenon beyond 5 years. Methods: From 345 patients with bipolar I or bipolar II disorder followed up for a mean (SD) of 13.7 (6.1) years as part of the National Institute of Mental Health Collaborative Depression Study, 89 (25.8%) were identified who, during 1 or more years of follow-up, manifested a pattern that met DSM-IV criteria for rapid cycling. These patients were compared with the remaining bipolar patients by demographics, overall affective morbidity, morbidity during specific treatment conditions, and the likelihood of suicidal behavior. Analyses assessed whether the use of tricyclic antidepressants for depressive symptoms was associated with the persistence of rapid cycling or with tendencies to switch from depressive to manic or hypomanic phases. Results: The 89 patients who showed a rapid cycling pattern were significantly more likely to have had an illness onset before 17 years of age and were more likely to make serious suicide attempts. In 4 of 5 cases, rapid cycling ended within 2 years of its onset. Resolutions were not associated with decreases in tricyclic antidepressant use. Throughout follow-up, patients prone to rapid cycling experienced more depressive morbidity than other bipolar patients, particularly when lithium carbonate was being used without tricyclic antidepressants. The use of these antidepressants was not more likely in the weeks preceding shifts from depression to mania or hypomania. Conclusions: These results indicate that bipolar patients who develop a rapid cycling pattern suffer substantial depressive morbidity and are at high risk for serious suicide attempts. These findings do not implicate tricyclic antidepressants or, by inference, serotonin reuptake inhibitors in the promotion of affective instability. Arch Gen Psychiatry. 2003;60: From the National Institute of Mental Health Collaborative Program on the Psychobiology of Depression Clinical Studies, Bethesda, Md. SEVERAL CONSIDERATIONS give importance to the phenomenon of rapid cycling. 1 In accord with the kindling hypothesis evoked to explain it, rapid cycling may reflect a sensitization produced by repeated episodes. 2,3 If, as suspected, such sensitization is often an enduring change, 3 the prognostic significance of rapid cycling should be long lasting as well. Some follow-up studies have supported this view, 4,5 but others have shown rapid cycling to be time limited. 6 Of perhaps greater concern to clinicians is the suggestion that antidepressants may both trigger and prolong rapid cycling The pattern often begins while antidepressants are being taken, and cycle lengths, the period from the beginning of one episode to the beginning of the next, have been seen to increase after antidepressant treatment was withdrawn. 7,8 These observations may be explained in other ways, however. Some investigators have noted that rapid cycling is likely to begin with depression 8 and that episodes that do begin with depression are substantially more likely to develop a polyphasic 9,10 or rapid cycling 6,11 course than are those that begin with mania. Because all studies of rapid cycling have described patients who sought treatment, those who did so during depressive phases were likely to be receiving antidepressants. Thus, depression may predispose both to antidepressant treatment and, independently, to phase shifting. Data from the National Institute of Mental Health Collaborative Depression Study 12 support this view. A regression analysis showed that the presence of depression preceding study intake, but not exposure to antidepressants, was an independent predictor of rapid cycling in the first year of follow-up. 6 A subsequent regression 914

2 analysis showed that depression during the first year of follow-up, but not antidepressant use in that year, predicted rapid cycling in the second year. 11 The conclusion that the discontinuation of antidepressants promotes the resolution of rapid cycling rests on 2 case series. In the first, 7 21 patients with rapid cycling were persuaded to discontinue antidepressant treatment, and 13 subsequently stabilized. The role of spontaneous stabilization could not be assessed, though, because the authors did not describe the course of patients who continued to take antidepressants. The second study 8 described cycle lengths in 9 rapid cyclers before and after discontinuation of tricyclic antidepressants (TCAs). Cycle lengths were longer during periods without TCAs. Because cycle length is timed from the beginning of one phase to the beginning of the next, it necessarily encompasses any intervening euthymic period. If depression portends further rapid cycling, as previous results have shown, and if antidepressants were more likely to have been discontinued during periods of euthymia than during periods of depression, as might be expected, then the association between antidepressant discontinuation and longer cycle lengths may have been artifactual. The presumption that antidepressants trigger or prolong rapid cycling suggests that patients who experience this phenomenon may be particularly liable to exhibit mania or hypomania while taking antidepressants. This may be especially true for TCAs. These were expressly implicated in the series by Wehr et al 8 and in a meta-analysis of placebo-controlled studies. 13 The latter showed a significantly higher likelihood of switching during TCA treatment than during treatment with selective serotonin reuptake inhibitors (SSRIs). If a liability to rapid cycling is inherent, then other features should distinguish bipolar patients who develop this pattern from those who do not. Of these, female sex has most consistently differentiated rapid cyclers from other bipolar patients. 6,7,14-16 In contrast, bipolar patients with and without rapid cycling have generally not differed by current age or by age at illness onset. However, none of the relevant reports have presented the distributions of the ages at which bipolar illness first appeared. Observations that rapid cycling or, more specifically, continuous cycling is very common among children and early adolescents who have bipolar illness 14,15 suggests 2 possibilities. Either childhood and adolescence have time-limited pathoplastic effects on bipolar illness that then resolve with adulthood, or very-earlyonset bipolar illness carries a liability to rapid cycling that is lifelong. If the latter is true, patients with very early onset should be overrepresented among those who show rapid cycling as adults. Because the identification of rapid cycling requires a careful mapping of episodes during a minimum time span of 1 year, a description of its natural course requires a lengthy observation period. In addition, conclusions as to the possible effects of treatment on the course of rapid cycling require that the members of a cohort be followed up prospectively regardless of whether they remain in treatment. The National Institute of Mental Health Collaborative Depression Study meets these specifications, and the following analysis uses data from it to address 4 questions: (1) Are very early onsets more common among adults who display rapid cycling? (2) How persistent is rapid cycling? (3) Are bipolar patients who are prone to this phenomenon more likely to experience manic or hypomanic syndromes while taking antidepressants than are other bipolar patients? (4) Is the resolution of rapid cycling associated with lower rates of antidepressant use? Our earlier report on the natural course of rapidcycling bipolar disorder began with the same cohort but described a substantially shorter observation period. 6 Moreover, because no fully operational definition of rapid cycling had then been published, we used one that seemed to best approximate the definitions provided in existing publications. The DSM-IV now lists clear criteria, but these differ in several respects from those we used earlier. Maj et al 16 showed that seemingly subtle variations in how rapid cycling is defined can have important effects on its prognostic significance. Given the wide use of the DSM-IV definition, it is important that the demographics and natural course of rapid cycling be described with the use of that definition. METHODS SUBJECTS Between 1978 and 1981, the National Institute of Mental Health Collaborative Program on the Psychobiology of Depression Clinical Studies recruited patients as they sought treatment for affective illness at any of 5 academic centers: Massachusetts General Hospital and Harvard University in Boston; Rush Presbyterian St Luke s Medical Center in Chicago, Ill; University of Iowa College of Medicine in Iowa City; New York State Psychiatric Institute and Columbia University in New York; and Washington University School of Medicine in St Louis, Mo. All patients were white, English speaking, and knowledgeable of their biological parents, and all met Research Diagnostic Criteria (RDC) 17 for major depressive disorder, schizoaffective disorder, or manic disorder. Each participant provided informed consent. The subjects described in the following analyses were limited to those who, at intake, met RDC for bipolar affective disorder, type I or II, or for schizoaffective disorder, manic type, other than the mainly schizophrenic subtype. Also included were subjects who met criteria for nonbipolar major depressive disorder at intake but who subsequently developed an episode of mania, hypomania, or schizoaffective disorder manic type. We used DSM-IV criteria to identify as rapid cyclers those patients who manifested, during any 52-week period of the follow-up, 4 or more mood episodes, each lasting 2 or more weeks, during which RDC for major depressive disorder, mania, hypomania, schizoaffective mania, or schizoaffective depression were met. Episodes or phases were separated either by switches to an opposite pole or by remissions lasting 8 or more weeks. A remission required 8 contiguous weeks with no more than 2 symptoms present to no more than a mild degree. PROCEDURES Intake assessments included the Schedule for Affective Disorders and Schizophrenia. 18 Ratings on this scale reflected information from direct patient interview, from informants, and from a review of medical records. 915

3 Raters recontacted probands at 6-month intervals for the first 5 years of follow-up and annually thereafter. They used information from direct patient interview and from medical records to complete the Longitudinal Interval Follow-up Evaluation (LIFE) 19 during the first 2 years, the LIFE II in years 2 to 5, and the Streamlined Longitudinal Interval Continuation Evaluation in the sixth year and beyond. The LIFE II and Streamlined Longitudinal Interval Continuation Evaluation differed from the LIFE in the omission of some psychosocial comparisons. These instruments tracked each RDC syndrome on a weekby-week basis. The interviewers identified change points for each syndrome active since the last assessment and derived symptom levels for each interval. The Psychiatric Status Rating used 6 levels for major depressive disorder, manic disorder, and schizoaffective disorder. A score of 1 indicated no symptoms and a 2 indicated the presence of no more than 1 or 2 symptoms to a mild degree. A 5 indicated a full syndrome and a 6, a relatively severe, full syndrome. Other RDC syndromes were rated on 3-point scales, and, for these, a 3 indicated the presence of a full syndrome. Eight consecutive weeks with no ratings greater than 2 indicated recovery from a specified syndrome, and a recurrence required the redevelopment of a full syndrome. The LIFE instruments also quantified psychosocial functioning in discrete areas of employment, household maintenance, academic work, interfamilial relationships, relationships with friends, and recreation. The LIFE Range of Impaired Functioning Tool condenses these various domain ratings to a single score with demonstrable reliability and validity. 20 Participation did not determine or influence treatment. Raters, though, systematically quantified all somatic treatment directed at mental disorders and, for each week, listed all individual drugs and their doses. 21 The following analyses entered treatment dichotomously as present or absent for any given week and separately considered lithium carbonate, valproate sodium, carbamazepine, TCAs, monoamine oxidase inhibitors, and SSRIs. ANALYSIS Survival analyses were used to quantify the persistence of rapid cycling patterns. These describe the time from the end of the first year in which rapid cycling was observed to the beginning of the first year in which rapid cycling was absent. Loss to follow-up was a censoring variable. In the absence of any widely accepted definition of rapid cycling resolution, we used 2 provisional ones. A narrow one specified the first year that encompassed no more than 1 episode and in which no switching occurred. The second, broader one identified the first year that encompassed no more than 2 episodes and no more than 1 switch. To determine whether patients who were prone to rapid cycling were more likely to develop manic or hypomanic symptoms while taking antidepressants, we focused on lithium and the TCAs. As will be shown, lithium was by far the dominant mood stabilizer prescribed for this cohort, and TCAs were used more than any other class of antidepressants. Moreover, the TCAs have been more strongly implicated than have the SSRIs in the provocation both of manic switches 13 and of rapid cycling. 22 Exposure to other mood stabilizers during this observation period was too limited to support analyses concerning their effects. The analyses compared those with any rapid cycling during follow-up with the remaining bipolar patients by the proportion of weeks in which manic or hypomanic symptoms coincided with antidepressant use. Thus, the number of weeks during which both Psychiatric Status Ratings for mania, hypomania, or schizoaffective mania were greater than 2, and any TCA treatment was ongoing, composed the numerator. The number of weeks with ongoing TCA treatment, regardless of coincident symptoms, composed the denominator. A daily TCA dose greater than an imipramine hydrochloride equivalent of 100 mg constituted ongoing TCA treatment. The likelihood of depressive symptoms during TCA treatment was similarly quantified. Depression was considered present when the Psychiatric Status Rating for major depressive disorder, schizoaffective depression, intermittent depression, or minor depression exceeded 2. To detect any association between the resolution of rapid cycling and changes in the use of TCAs, analyses focused on the use of TCAs in the final 52 weeks of the first rapidcycling period observed during follow-up and on the subsequent 52 weeks representing the first year in which rapid cycling was no longer present. We chose not to simply compare the number of weeks of TCA use in the 2 periods because a resolution of rapid cycling implies less psychopathology generally and, because less morbidity leads to less treatment, an association between the resolution of rapid cycling and a decreased use of TCAs may have been artifactual. Instead, the analyses considered only weeks during which depressive symptoms were present and compared the 2 periods by the proportion of those weeks during which TCAs were prescribed. Those who developed rapid cycling during follow-up were compared with those who did not by means of 2 tests for categorical variables and t tests for continuous variables. A 2-tailed level of.05 was used for each statistical test. RESULTS Of 345 patients who met inclusion criteria for this analysis, 89 (25.8%) met DSM-IV criteria for rapid cycling during at least 1 of the years of prospective observation. Sixtynine (78%) of these patients displayed rapid cycling in the first year of follow-up, and another 11 (12%) began rapid cycling in the second year. Only 1 patient had the onset of rapid cycling later than the eighth year of followup. The modal number switches in the first year of rapid cycling was 4 (40 subjects); 33 (37%) had fewer than 4 switches, and only 1 (1%) had 10 or more. Forty-six (52%) had no periods of remission (2 or more symptom-free months) during that year. This group was followed up for a mean (SD) of 15.2 (4.9) years and the remaining bipolar patients were followed up for a shorter mean (SD) of 13.2 (6.4) years (t 343 = 2.54, P=.01). More than two thirds of those who developed a rapid-cycling pattern were female, while the sex ratio for the remaining patients was nearly even (Table 1). Bipolar II disorder was not more frequent among those prone to rapid-cycling patterns. These patients were somewhat more likely to be in a depressed or mixed state at the beginning of follow-up and had had, on average, more than twice as many previous episodes of affective disorder. Although the 2 groups were similar in ages at intake, those who developed rapid cycling were significantly younger when they first met criteria for an affective disorder. Distributions of age at onset showed that onsets in childhood and in early puberty accounted for this difference (Figure 1). With the threshold for earlyonset bipolar disorder proposed by Geller et al, (30%) of the rapid-cycling patients and 40 (15.6%) of the remaining bipolar patients had an onset at age 16 years or younger ( 2 1=9.75, P.005). 916

4 Table 1. Baseline Measures Any Rapid Cycling During Follow-up (n = 89) No Rapid Cycling During Follow-up (n = 256) Sex, No. (%) F* 62 (69.7) 137 (53.5) Inpatient, No. (%) 71 (79.8) 214 (83.6) Age at intake, mean (SD), y 36.0 (11.8) 36.1 (13.0) Age at first outpatient contact, 23.4 (8.6) 25.3 (10.0) mean (SD), y Age at first episode of affective 21.6 (8.9) 24.4 (10.6) disorder, mean (SD), y No. of previous episodes, 26.5 (49.6) 11.1 (29.0) mean (SD) Bipolar I, No. (%) 66 (74.2) 181 (70.7) Bipolar II, No. (%) 29 (32.6) 89 (34.8) Polarity at intake, No. (%) Depressed 57 (64.0) 140 (54.7) Manic 22 (24.7) 97 (37.9) Mixed 9 (10.1) 16 (6.2) * 2 1 = 7.0, P =.008. t 343 = 2.2, P =.03. t 343 = 3.54, P.006. Data were missing for 1 subject in the group with any rapid cycling during follow-up and for 3 subjects in the group with no rapid cycling during follow-up. Among rapid-cycling patients, the 27 with onsets before age 17 years did not differ from the remaining 62 by liability to switching or by the quantity of affective morbidity during follow-up. Rapid-cycling patients with early onsets and those with later onsets experienced means (SDs) of 2.5 (1.3) and 2.5 (1.5) switches per year, respectively. The respective mean (SD) proportions of follow-up weeks with depressive symptoms were 0.48 (0.26) and 0.39 (0.27) (t 87 =1.48, P=.14). The mean (SD) proportions of weeks with manic symptoms were 0.11 (0.11) and (0.14), respectively (t 87 = 0.93, P=.36). Over all weeks of follow-up, those who at any time showed a rapid-cycling course experienced depressive symptoms for 0.42 (SD, 0.27) of the follow-up weeks, a significantly higher proportion than the 0.30 (SD, 0.30) seen among the remaining bipolar patients (t 343 = 3.31, P=.001). Rapid-cycling patients also tended to experience more weeks with manic or hypomanic symptoms. Such symptoms were present during 0.13 (SD, 0.13) of the follow-up weeks among rapid cyclers and during 0.09 (0.16) of the weeks among nonrapid cyclers (t 343 = 1.91, P=.06). The mean ratio of weeks depressed to weeks manic or hypomanic was 11.1 (SD, 18.7) for rapid cyclers and 32.6 (SD, 97.9) for non rapid cyclers (t 343 =2.05, P=.04). The higher symptom levels that characterized patients prone to rapid cycling were reflected in greater overall functional impairment. The mean (SD) LIFE Range of Impaired Functioning Tool scores for these patients, and for the remaining bipolar patients, were 11.9 (2.4) and 10.7 (2.7), respectively (t 343 = 3.56, P.001). Only 20 patients manifested rapid cycling for 2 or more consecutive years. Of these, 13 displayed the pattern for only 2 years, 4 for only 3 years, 3 for 4 years. The 20 with 2 or more consecutive years of rapid cycling did not differ significantly from the remaining 69 by any of the variables listed in Table 1. % With Onset Age at Onset, y Figure 1. Age at onset in bipolar affective disorder. % With Persistence of Rapid Cycling Rapid Cycling (n = 89) No Rapid Cycling (n = 256) Weeks From End of First Year of Rapid Cycling Narrow Broad Figure 2. The resolution of rapid cycling by definition, n=83. The narrow definition specified a first year that encompassed no more than 1 episode and in which no switching occurred. The broad definition identified a first year that encompassed no more than 2 episodes and no more than 1 switch. More than 4 in 5 episodes of rapid cycling were observed to end whether the threshold for this required fewer than 3 episodes or 2 switches in a 1-year period, or fewer than 2 episodes and no switches (Figure 2). Six subjects had no follow-up beyond their first year of rapid cycling and were therefore omitted from this analysis. According to Kaplan-Meier estimates from the survival analyses, the median number of weeks from the end of the first year in which rapid cycling was observed, to the beginning of the first year of stabilization, was 6 for the first definition and 23 for the second, stricter definition. Of patients whose rapid-cycling pattern was observed to resolve according to the broader definition, 64 had at least some weeks of depression during the year of stabilization. These patients were no less likely to have taken TCAs while depressed in the period after the resolution of rapid cycling than they had in the preceding year when rapid cycling was ongoing. Tricyclic antidepressants were prescribed for 32.1% (SD, 42.6%) of the 2035 person-weeks with depression observed during the first year of resolution; the corresponding percentage for the preceding year was 28.8% (SD, 34.8%) during 2003 person-weeks with depression. Nor did these 2 years differ 917

5 Table 2. Treatment Conditions and Morbidity During Follow-up Treatment % of Weeks of Each Treatment, Mean (SD) Any Rapid Cycling (n = 89) No Rapid Cycling (n = 256) Lithium carbonate 45.0 (37.1) 45.2 (40.0) TCAs 24.0 (24.0) 20.0 (23.0) MAOIs 4.3 (15.4) 3.9 (12.6) SSRIs* 7.2 (11.9) 4.1 (10.0) Lithium with TCA 13.7 (20.9) 7.7 (16.8) Lithium without TCA 31.2 (31.5) 37.4 (38.1) TCA without lithium 5.5 (9.0) 5.7 (12.2) Carbamazepine 5.1 (13.3) 3.4 (10.9) Valproate sodium 2.6 (5.5) 1.1 (4.3) Abbreviations: MAOIs, monoamine oxidase inhibitors; SSRIs, selective serotonin reuptake inhibitors; TCAs, tricyclic antidepressants. *t 343 = 2.43, P =.02. t 343 = 2.72, P =.007. t 343 = 2.59, P =.01. Table 3. Manic or Hypomanic and Depressive Symptoms During Lithium Carbonate Therapy With or Without TCAs Any Rapid Cycling No Rapid Cycling Person-weeks of treatment with lithium + TCA % Of weeks with mania or hypomania, 17.3 (27.1) 9.5 (22.6) mean (SD) % Of weeks with depression, mean (SD) 56.2 (35.8) 61.8 (37.5) Person-weeks of treatment with lithium without TCA % Of weeks with mania or hypomania, 21.5 (26.0) 15.9 (25.6) mean (SD) % Of weeks with depression, mean (SD)* 46.8 (36.2) 28.6 (36.4) Abbreviation: TCAs, tricyclic antidepressants. *t 286 = 3.80, P.001. in the percentage of weeks during which mood stabilizers (lithium, carbamazepine, or valproate sodium) were prescribed. These were 48.7% (SD, 45.8%) for the rapidcycling year and 49.3% (SD, 47.3%) for the following year. Bipolar patients who displayed any rapid cycling took SSRIs, lithium with TCAs, and valproate for greater proportions of the follow-up weeks than did those who showed no rapid cycling (Table 2). Otherwise, the groups had similar treatment exposures. During periods when lithium was being taken without TCAs, patients prone to rapid cycling were significantly more likely to be experiencing depressive symptoms than were other bipolar patients (Table 3). During weeks when lithium was taken with a TCA, patients in the 2 groups had similar likelihoods of depressive symptoms. Patients prone to rapid cycling were somewhat more likely than other bipolar patients to have manic or hypomanic symptoms in any given week during lithium treatment, but this was true both in the presence and in the absence of TCAs. We considered whether patients prone to rapid cycling were more likely to have been taking a TCA during those weeks immediately preceding a switch from depression to mania or hypomania than they were during Table 4. Switches From Depression to Mania or Hypomania and TCA Use Among 89 Patients Prone to Rapid Cycling Weeks Preceding at Switch All Other Weeks With Depression Total person-weeks observed No. (%) of weeks with any TCA 136 (28.8) 7140 (29.8) No. (%) of weeks with any lithium 95 (20.1) 4164 (17.4) carbonate + TCA No. (%) of weeks with any lithium 143 (30.3) 6983 (29.1) without TCA No. (%) of weeks with any TCA without lithium* 41 (8.7) 2976 (12.4) Abbreviation: TCA, tricyclic antidepressant. * 2 1 = 6.6, P =.01. Table 5. Suicidal Behaviors Before Intake and During Follow-up Any Rapid Cycling (n = 89) No Rapid Cycling (n = 256) Before intake, No. (%) Any attempt* 51 (57.3) 85 (33.2) Any attempt of high intent 31 (34.8) 38 (14.8) Any attempt of high lethality 36 (40.4) 35 (13.7) After intake, No. (%) Completed suicide 3 (3.4) 11 (4.3) Any attempt 46 (51.7) 70 (27.3) Any attempt of high intent 27 (30.3) 33 (12.9) Any attempt of high lethality 30 (33.7) 36 (14.1) * 2 1 = 16.1, P = 16.5, P = 29.0, P = 17.5, P = 14.0, P = 16.5, P.001. other weeks. The results did not implicate TCAs in these switches (Table 4). These patients were no more likely to have been taking a TCA, with or without lithium, during the weeks preceding a switch from depression to mania than they were during any other week when depression was present. Although patients with rapid cycling were not more likely to commit suicide (Table 5), they were more likely than the remaining bipolar patients to have made attempts before study intake and to make them during the follow-up period. Group differences were particularly large for attempts rated as being high in intent or in potential medical lethality. Sex is a well-known risk factor for attempted suicide, and females were overrepresented in the rapidcycling group. A logistic regression analysis with sex and the presence or absence of rapid cycling as independent variables showed that both were significantly associated with the likelihood of suicide attempts during followup. The odds ratios were 2.0 (95% confidence interval, ) for female sex (Wald 2 =7.5, P=.006) and 2.6 (95% confidence interval, ) for rapid cycling (Wald 2 =13.8, P.001). 918

6 National Institute of Mental Health Collaborative Program on the Psychobiology of Depression Clinical Studies The National Institute of Mental Health Collaborative Program on the Psychobiology of Depression Clinical Studies was conducted with the participation of the following investigators: M. B. Keller, MD (Chairperson, Providence, RI); W. Coryell, MD (Co-chairperson, Iowa City, Iowa); T. I. Mueller, MD, D. Solomon, MD (Providence); J. Fawcett, MD, W. A. Scheftner, MD (Chicago, Ill); W. Coryell, MD, J. Haley (Iowa City); J. Endicott, PhD, A. C. Leon, PhD, J. Loth, MSW (New York, NY); J. Rice, PhD, T. Reich, MD (St Louis, Mo). Other contributors include H. S. Akiskal, MD, N. C. Andreasen, MD, PhD, P. J. Clayton, MD, J. Croughan, MD, R. M. A. Hirschfeld, MD, L. Judd, MD, M. M. Katz, PhD, P. W. Lavori, PhD, J. D. Maser, PhD, M. T. Shea, PhD, R. L. Spitzer, MD, M. A. Young, PhD. Deceased: G. L. Klerman, MD, E. Robins, MD, R. W. Shapiro, MD, G. Winokur, MD. COMMENT Bipolar patients who developed a rapid-cycling pattern at any time during follow-up were more likely than those who did not to have had onsets before 17 years of age, were more likely to make serious suicide attempts during follow-up, and were much more likely to have depressive symptoms during any given week. The greater depressive morbidity experienced by patients prone to rapid cycling was reflected in their increased exposure to TCAs while taking lithium. They were somewhat more likely to experience mania or hypomania while taking lithium, but this was true whether or not TCAs were also present. Moreover, TCAs were no more often present in weeks preceding shifts from depression to mania or hypomania than in other weeks when depression was present. Although the course of patients who showed rapid cycling patterns differed in several ways from the course of bipolar patients without it, rapid cycling in itself was a transient phenomenon in the majority of cases. It is likely, though, that the persistence of rapid cycling when identified prospectively, as in this study, will be less than that seen in samples selected for a current rapid-cycling pattern, as the latter method will oversample patients whose rapid cycling is truly chronic. It is notable, then, that the only other report to quantify the persistence of rapid cycling during a lengthy follow-up found that a similar 19% continued to show this pattern. 4 On the other hand, the current sample was recruited at tertiary care centers and the patients described by Maj et al 4 all attended a clinic specializing in affective disorders. As such, patients with refractory disease were probably overrepresented in both samples relative to patients typically seen by private practitioners. These analyses found no association between the cessation of rapid-cycling patterns and any decrease in the use of TCAs to treat depressive symptoms during lithium maintenance. Notably, another recent report found that the discontinuation of antidepressants in a group of bipolar patients was associated, not with a decrease in cycling, but with an increase in depressive morbidity. 24 Bipolar patients who manifested rapid cycling resembled those who did not by the proportion of follow-up weeks during which they were treated with lithium and/or TCAs. The proportion of weeks with SSRI treatment was low in both groups but significantly higher for those with rapid cycling. This may have reflected the fact that the somewhat longer observation period for this group extended further into the period when SSRIs were displacing TCAs. The numbers of weeks observed during SSRI treatment were quite small for both groups, though, and this precluded efforts to determine their effects on affective stability. While taking lithium, patients prone to rapid cycling were not significantly more likely to manifest manic or hypomanic symptoms when TCAs were also present than were patients who showed no such liability. Nor was the use of TCAs associated with an increased risk for switching in this affectively unstable group. Because TCAs have been more strongly implicated than other antidepressants in the promotion of such switches, these results probably apply as well to the newer antidepressants. This has important implications for the management of bipolar depression generally and for rapid cycling in particular. Switches did indeed occur while TCAs were being used, but they occurred no less frequently when TCAs were not in use. These findings are quite consistent with those of studies with placebo control. Three groups have randomly assigned bipolar patients to treatment with lithium with placebo or lithium with TCAs for prophylaxis, and none found manic episodes to be significantly more likely among the latter patients Similarly, a naturalistic study of outpatients with bipolar I maintained on a regimen of mood stabilizers (predominantly lithium) found that depressive episodes in which TCAs were used were no more likely to end in switches than were depressive episodes managed without antidepressants. 28 Rapid-cycling patients were more than twice as likely as other bipolar patients to have a history of severe suicide attempts at the beginning of follow-up, and to make further serious attempts during follow-up. This is consistent with our finding that rapid-cycling patients were depressed for significantly greater proportions of the follow-up and, presumably, experienced more depression before they entered follow-up. The abrupt transitions from mania or hypomania to depression that many of these patients experience may further predispose to suicidality. It is notable, though, that no trend emerged to indicate that rapid cyclers were at greater risk for completed suicide despite the large group differences in serious attempt rates. A clear association between suicidality and rapid cycling has not emerged in earlier descriptions, although Bauer et al 29 and Maj et al 4 described strong trends in this direction. Wu and Dunner 30 specifically sought an association between rapid cycling and suicide attempts in a 919

7 chart review of bipolar patients and found none. Their sample was drawn from a well-known specialty clinic and contained a large proportion of patients with treatmentrefractory disease, however. Whether such an association is characteristic of more representative samples awaits further study. Early onsets were overrepresented among adult patients prone to rapid cycling, and yet rapid-cycling patients with an onset of bipolar illness before the age of 17 years resembled other rapid-cycling patients in their propensity to switching and in overall manic or depressive morbidity over time. Although these data cannot determine the direct effects of prepubertal and early pubertal periods on the propensity to rapid cycling, they do indicate that the very early onset of bipolar disorder may convey a lifelong propensity to this phenomenon. Submitted for publication May 7, 2002; final revision received January 24, 2003; accepted January 27, This study was supported by grants MH25416 and MH51324 from the National Institute of Mental Health, Bethesda, Md. This article has been reviewed by the Publication Committee of the Collaborative Depression Study and has its endorsement. Corresponding author: William Coryell, MD, Psychiatry Research MEB, Room 2-205, University of Iowa Health Care, Iowa City, IA ( REFERENCES 1. Dunner DL, Fieve RR. Clinical factors in lithium carbonate prophylaxis failure. Arch Gen Psychiatry. 1974;30: Post RM, Robinson DR, Ballenger JC. Conditioning and sensitization in the longitudinal course of affective illness. Br J Psychiatry. 1986;149: Post RM. Transduction of psychosocial stress into the neurobiology of recurrent affective disorder. Am J Psychiatry. 1992;149: Maj M, Magliano L, Pirozzi R, Marasco C, Guarneri M. Validity of rapid cycling as a course specifier for bipolar disorder. Am J Psychiatry. 1994;151: Goldberg JF, Harrow M. Kindling in bipolar disorders: a longitudinal follow-up study. Biol Psychiatry. 1994;35: Coryell W, Endicott J, Keller M. Rapidly cycling affective disorder: demographics, diagnosis, family history, and course. Arch Gen Psychiatry. 1992;49: Kukopulos A, Reginaldi D, Laddomada P, Floris G, Serra G, Tondo L. Course of the manic depressive cycle and changes caused by treatment. Pharmakopsychiatr Neuropsychopharmakol. 1980;13: Wehr TA, Sack D, Rosenthal N, Cowdry R. Rapid cycling affective disorder: contributing factors and treatment responses in fifty-one patients. Am J Psychiatry. 1988;145: Turvey CL, Coryell WH, Arndt S, Solomon DA, Leon AC, Endicott J, Mueller T, Keller M, Akiskal H. Polarity sequence, depression, and chronicity in bipolar I disorder. J Nerv Ment Dis. 1999;187: Baldessarini RJ, Tondo L, Floris G, Hennen J. Effects of rapid cycling on response to lithium maintenance treatment in 360 bipolar I and II disorder patients. J Affect Disord. 2000;61: Coryell W. Can antidepressants induce rapid cycling? Arch Gen Psychiatry. 1993; 50: Keller MB, Lavori PW, Coryell W, et al. Differential outcome of pure manic, mixed/ cycling, and pure depressive episodes in patients with bipolar illness. JAMA. 1986; 255: Peet M. Induction of mania with selective serotonin reuptake inhibitors and tricyclic antidepressants. Br J Psychiatry. 1994;164: Geller B, Sun K, Zimmerman B, Luby J, Frazier J, Williams M. Complex and rapid cycling in bipolar children and adolescents: a preliminary study. J Affect Disord. 1995;34: Geller B, Craney JL, Bolhofner K, delbello MP, Williams M, Zimmerman B. Oneyear recovery and relapse rates of children with a prepubertal and early adolescent bipolar disorder phenotype. Am J Psychiatry. 2001;158: Maj M, Trozzi R, Formacola AMR, Tortorella A. Reliability and validity of four alternative definitions of rapid cycling bipolar disorder. Am J Psychiatry. 1999; 156: Spitzer RL, Endicott J, Robins E. Research Diagnostic Criteria: rationale and reliability. Arch Gen Psychiatry. 1978;35: Endicott J, Spitzer AL. Kindling in bipolar disorders: a longitudinal follow-up study. Arch Gen Psychiatry. 1978;35: Keller MB, Lavori PW, Friedman B, Nielsen E, Endicott J, McDonald-Scott P, Andreasen NC. The Longitudinal Interval Follow-up Evaluation. Arch Gen Psychiatry. 1987;44: Leon AC, Solomon DA, Mueller TI, Turvey CL, Endicott J, Keller MB. The range of impaired functioning tool (LIFE-RIFT): a brief measure of functional impairment. Psychol Med. 1999;29: Keller MB, Lavori PW, Klerman GL, Andreasen NC, Endicott J, Coryell W, Fawcett J, Rice JP, Hirschfeld RM. Low levels and lack of predictors of somatotherapy and psychotherapy received by depressed patients. Arch Gen Psychiatry. 1986;43: Wehr TA, Goodwin FK. Rapid cycling in manic depressives induced by tricyclic antidepressants. Arch Gen Psychiatry. 1979;36: Geller B, Zimerman B, Williams M, Bolhofner K, Craney JL, Delbello MP, Soutullo CA. Six-month stability and outcome of a prepubertal and early adolescent bipolar disorder phenotype. J Child Adolesc Psychopharmacol. 2000;10: Altshuler LL, Kiriakos L, Calcagno J, Goodman R, Gitlin M, Frye M, Mintz J. The impact of antidepressant discontinuation versus antidepressant continuation on 1-year risk for relapse of bipolar depression: a retrospective chart review. J Clin Psychiatry. 2001;62: Quitkin FM, Kane J, Rifkin A, Ramos-Lorenzi JR, Nayak DV. Prophylactic lithium carbonate with and without imipramine for bipolar I patients: a double-blind study. Arch Gen Psychiatry. 1981;38: Prien RF, Kupfer DJ, Mansky PA, Small JG, Tuason VB, Voss CB, Johnson WE. Drug therapy in the prevention of recurrences in unipolar and bipolar affective disorders: report of the NIMH Collaborative Study Group comparing lithium carbonate, imipramine, and a lithium carbonate imipramine combination. Arch Gen Psychiatry. 1984;41: Nemeroff CB, Evans DL, Gyulai L, Sachs GS, Bowden CL, Gergel IP, Oakes R, Pitts CD. Double-blind, placebo-controlled comparison of imipramine and paroxetine in the treatment of bipolar depression. Am J Psychiatry. 2001;158: Boerlin HL, Gitlin MJ, Zoellner LA, Hammen CL. Bipolar depression and antidepressant-induced mania: a naturalistic study. J Clin Psychiatry. 1998;59: Bauer MS, Calabrese J, Dunner DL, Post R, Whybrow PC, Gyulai L, Tay LK, Youkin SR, Bynum D, Lavori P, Price RA. Multisite data reanalysis of the validity of rapid cycling as a course modified for bipolar disorder in DSM-IV. Am J Psychiatry. 1994;151: Wu LH, Dunner DL. Suicide attempts in rapid cycling bipolar disorder patients. J Affect Disord. 1993;29: