Drug Treatment Of Mania: Assessing the Evidence

By | February 8, 2015

To prove that a drug is efficacious in a psychiatric condition, it is essential to show that it is superior to placebo, by conducting randomized double-blind placebo-controlled trials. The challenges of conducting such trials in mania have only been met in recent years, in the course of developing novel anticonvulsant and atypical antipsychotic treatments. These recent trials are therefore providing answers to questions that have long remained unresolved about the treatment of mania. Analysis of the results of these trials requires attention not only to the statistical significance of differences in special rating scales, but also to the size of the effect, and to the generalizability of results, which have been derived from highly selected patients in clinical trial centers, to other groups of patients with mania in routine practice. It is also important to consider how dropouts from the studies may have biased the interpretation of results.

High placebo response rates (with the exception of the Pope et al. study show how vital placebo-controlled studies are in identifying a drug’s efficacy in a way that limits any potential bias. Table: Drugs Shown to be Superior to Placebo as Monotherapy in Mania in RCTs lists the drugs that have been proved superior to placebo as monotherapy in such trials. To receive a license to market a drug for mania, most authorities, including the European Medicines Agency, require two trials performed at independent centers.

Table: Drugs Shown to be Superior to Placebo as Monotherapy in Mania in RCTs

Drug Number of trials
Valproate 2
Lithium 4
Haloperidol 2
Olanzapine 2
Risperidone 3
Quetiapine 2
Ziprasidone 2
Aripiprazole 2
Carbamazepine 2

Lithium, Valproate, or Antipsychotics for Mania

The first drug to be proved efficacious in such trials was valproate. Although Table 3 shows that the drug with most trials (not all published) proving efficacy is lithium, this drug is not usually sufficiently rapid in onset to be useful as mono-therapy. Although haloperidol has been the favorite drug of clinicians for treating mania, it is only in the course of comparative trials with risperidone and quetiapine, that haloperidol has been proved conclusively to be efficacious. Most evidence for efficacy in mania now concerns the atypical antipsychotics, olanzapine, risperidone, quetiapine, ziprasidone, and aripiprazole. Questions arise about the relative efficacy of these atypical antipsychotics compared either with haloperidol or with valproate, and whether they should be used initially as monotherapy or combined with valproate or lithium. Also, since treatment often needs to commence with rapid tranquillization, and only two atypicals can be given intramuscularly (recently olanzapine and in some countries ziprasidone), haloperidol remains widely used despite its propensity to cause unpleasant extrapyramidal side effects.

Placebo-Controlled RCTs of Atypical Antipsychotics

The published trials (some as yet presented only in Posters at meetings) are summarized in Table Monotherapy with Atypical Antipsychotics in Mania: Placebo-Controlled Parallel-Group Randomized Trials, using a “Number-Needed-to-Treat” (NNT) analysis. NNT is calculated by dividing the difference in response rate between active drug and placebo into 100 and correcting to the next highest integer. It represents the number of patients who must be treated in order for one patient to achieve the defined response — usually a 50% reduction in score on a scale such as the 11-item Young Mania Rating Scale (YMRS) — as a result of the pharmacological effect of the drug. NNT thus provides a measure of the size of effect that can be expected of the drug in a clinical situation. For a drug to be useful monotherapy as a first-line treatment in a common and severe disorder such as mania, the NNT for 50% improvement in severity should be in the order of 2-4.

Most studies were of 3 or 4 weeks duration and there was a placebo response rate of 19-43%, reflecting the effects of a variety of possible non-specific factors such as hospitalization, extra medication with benzodiazepines or chloral allowed during the first 10 days, and bias in the raters. Dropout rates for inefficacy ranged from 10% to 69%o on placebo, and from 7% to 47% on active drug. Dropouts through adverse events (including suspected side effects) ranged from 1.5% to 10% on placebo, and from 0%o to 11% on atypical antipsychotic; on haloperidol dropouts through lack of efficacy were 7% and 35% and through adverse events 3% and 10%. Total dropout rates ranged from 15% to 79% on placebo, from 11% to 58% on atypical antipsychotic, and were 10% and 45% on haloperidol and 32% on lithium.

Olanzapine. Olanzapine was the first of the atypical antipsychotics to be proved efficacious in mania in trials designed by Tohen et al.. Patients entering the studies had mean mania scores of 29 on the Young Mania Rating Scale (maximum possible score 46). Patients with mixed mania could also be included. The starting doses in the two studies were 10 and 15mg, and the mean modal doses were 14.9 and 16.4 mg/day. The commonest side effects were somnolence (22% more than placebo), dry mouth (15% more), dizziness (12% more), weakness (9% more), and weight gain.

Risperidone. In the trial of Hirschfeld et al., risperidone was increased gradually over 4 days to a maximum of 6 mg/day. The NNT was 6 (95% confidence interval. Side effects on risperidone were somnolence (28% vs. 7% on placebo), hyperkinesias (Parkinsonism) in 16% vs. 5%.

The study by Khanna et al., conducted in India, was distinguished by relatively high Young Mania Rating Scale scores on entry (about 10 points higher than in most other studies of atypical antipsychotics in mania), and by a high rate of study completion on risperidone with low dropout rates particularly for lack of efficacy. The mean modal dose in this study was higher (5.4 mg/day). The rate of response on risperidone was highest in this study and the NNT was impressively low at 3 (95% confidence interval. Extrapyramidal side effects occurred in 35% of those on risperidone, and 6% on placebo.

The study by Eerdekens et al. had a slower dosing schedule reaching a maximum of 6 mg/day by day 5. This study also had a haloperidol comparator group with a mean modal dose of 8 mg/day (see below). Both active drugs were effective compared with placebo from day 7. By day 21, the NNT for 50% improvement was 7 (95%o confidence interval for risperidone and similar for haloperidol at 8 (95%o confidence interval. Side effects on risperidone included extrapyramidal symptoms (17% compared with 40% on haloperidol).

In the studies of Khanna et al. and of Eerdekens et al., the improvement on active treatments was continuing to develop during the third week of treatment; the extension to 12 weeks in the Eerdekens study demonstrated further improvement in Young Mania Rating Scale scores with active treatment up to 12 weeks, suggesting that the dose initially administered may not have been sufficient to achieve maximum improvement. On the other hand, in the study of Hirschfeld et al., the improvement appeared to be complete within 2 to 3 weeks.

Risperidone had been described as worsening or precipitating mania in predisposed patients, mainly those with schizophrenia, schizoaffective disorder, or organic brain disease. No evidence of such exacerbation of mania was found in these placebo-controlled trials of risperidone.

Quetiapine. The two trials of monotherapy enrolled patients with relatively high scores on the Young Mania Rating Scale (average 33 points); patients with mixed mania were excluded. Significant efficacy at 3 weeks was observed in one study and in the combined analysis. In the 12-week extension, both trials showed significant efficacy but the dropout rate on placebo was very high. Patients who responded to quetiapine were usually receiving 600 mg/day or more. The dose was increased toward this amount over 5 days and then to a maximum of 800 mg/day. The commonest side effects were somnolence, dry mouth, weight gain, and dizziness.

Ziprasidone and Aripiprazole. The results of trials of these drugs are also shown in Table 4. In common with most of the other trials, these had high dropout rates on placebo. The dropout rate on aripiprazole was somewhat higher than on other drugs. Number needed to treat was slightly higher with ziprasidone.

Placebo-Controlled Monotherapy Trials of Haloperidol in Mania

Two monotherapy studies have included haloperidol as an active comparator, the risperidone study of Eerdekens et al., and the quetiapine study of Brecher et al.

In the study of Eerdekens et al, the haloperidol dose started at 4 mg/day and was adjusted to 2-12 mg/day by day 5. The timecourse of improvement was similar to that with risperidone, and by day 21, the NNT for 50% improvement was 8 (95%o confidence interval. This is far larger than one would expect with the most commonly used antimanic drug of the previous decade; this might be either because the mean modal dose of haloperidol was only 8 mg/day, or because the patients in the trial were in some ways not typical of routine clinic patients and were more resistant to treatment.

A comparator group on haloperidol (up to 8 mg/day) was also included in the study by Brecher et al. of quetiapine (up to 800 mg/day) and placebo, analyzed at 3 and 12 weeks. At 3 weeks, the response rate (50% reduction in Young Mania Rating Scale score) on haloperidol, on a mean dose of only 5.2 mg/day, was 55% compared with 35% on placebo, giving an NNT of 5 (95% confidence interval. There were more dropouts on placebo than on haloperidol or quetiapine, so that the analysis using last observations carried forward was biased in favor of the active drugs, and especially so after 3 weeks when more patients on placebo or haloperidol than on quetiapine dropped out. By 12 weeks, the response rate on haloperidol was 70% and on placebo 39%, giving an NNT of 4. Side effects in the form of extrapyramidal symptoms were much more common on haloperidol (59.6%) than on placebo (15.8%), as was akathisia (33.3% on haloperidol and 5.9% on placebo). Somnolence occurred more often with haloperidol (9.1%) than placebo (5%).

Patterns of Symptom Improvement: Sedative, Antipsychotic, or Antimanicl

It had been suggested that antipsychotics owe their effects in mania either to non-specific sedation (that is making the person drowsy or asleep), or to combating psychotic symptoms. However, this view fails to recognize that non-sedative dopamine-blocking drugs can improve mania. In all studies of olanzapine and risperidone and in the combined analysis of quetiapine studies, the improvement in mania occurred in patients with or without psychotic symptoms. When individual items of the mania rating scale were analyzed, drug treatment (with olanzapine, quetiapine, and probably the other atypicals) improved the whole range of symptoms (including elation, flight of ideas, grandiosity, sexual interest, irritability, aggression, and general appearance, as well as the items most sensitive to sedation: insomnia, overactivity, and pressure of speech).

These findings lead to the inevitable conclusion that the drugs are not just antipsychotic, and in some cases, sedative, but antimanic.

Depression in Mania

Depressive symptoms are very common during mania, and if amounting to a major depressive syndrome, the condition is classified as mixed mania according to DSM-IV. However, at least 12 forms of bipolar mixed states have been described and are likely to respond differently to treatments.

It has been suggested, but never proved, that classical antipsychotics may worsen or induce depression, apart from their obvious extrapyramidal side effects. In the trials of atypical antipsychotics, the changes in symptoms of depression have usually been monitored. Three drugs have been demonstrated to improve depressive symptoms alongside the improvement in mania; these are olanzapine, risperidone, and aripiprazole. Olanzapine improved depression scores more than placebo. In the study of Khanna et al., depression scores were already low at the start, but improved further, the change being greater on risperidone than placebo even by day 3. In the study of Eerdekens et al., Montgomery-Asberg Depression Rating Scale depression scores, low at the start, fell more on risperidone than on placebo from week 1, and on haloperidol only at week 2.

In the study by Brecher et al., depression scores (MADRS) improved by day 21 on both quetiapine and haloperidol, more than on placebo. In trials that permitted inclusion of mixed mania, it has been shown that this improves with treatment in the case of olanzapine.

Comparative RCTs ofAntipsychotics in Mania Without Placebo

In a comparative trial in mania, in which additional lorazepam was permitted, risperidone showed similar efficacy to haloperidol or lithium.

In the largest randomized comparative study of haloperidol, it was compared with olanzapine over 6 and 12 weeks. Among patients on haloperidol (up to 15 mg/day, mean modal dose at sixth week 7 mg/day), the proportion responding (50% reduction in Young Mania Rating Scale score) at 6 weeks was 74%; the proportion showing syndromal remission (according to DSM-IV) was 44%, a figure similar to that found in consecutive admissions for mania by Rifkin et al.. Improvement in mania scores (YMRS) was greater for haloperidol than for olanzapine at 6 weeks, but not different at 12 weeks. In patients with low levels of depressive symptoms at commencement on haloperidol, a total of 16.8% switched into depression within 12 weeks. However, as there was no placebo group, it is not known whether this represents the natural history of the patients’ mood cycles, perhaps accelerated by effective treatment of mania, or some additional depressant effect of haloperidol. The switch rate among patients on olanzapine was lower at 6 weeks, but at 12 weeks, the difference was not significant.

In the study by Brecher et al., the switch rates into depression over 12 weeks were similar for haloperidol (8.1%) and placebo (8.9%), and tended to be lower for quetiapine (2.9%).

Thus, both olanzapine and quetiapine show trends to produce a lower switch rate into depression than haloperidol. However, both drugs (in the doses used) seemed also to lead to slower improvement in mania than haloperidol.

Extrapyramidal symptoms occur to a much less extent with olanzapine or quetiapine than with haloperidol. In the largest comparative trial, treatment-emergent akathisia was observed in 40% on haloperidol and 10% on olanzapine, dystonia in 6.8% and 1.3%, and Parkinsonism in 54% and 13%, respectively.

Trials of Atypical Antipsychotic Vs. Valproate in Mania

A study comparing oral haloperidol (0.2 mg/kg/day) with high initial doses of valproate semisodium (20 mg/kg/day) in psychotic mania found similar timecourse of improvement with both the drugs. However, this finding may not be general-izable, as the response to haloperidol was unusually slow. There have been two studies comparing olanzapine with valproate (as valproate semisodium or divalproex) in mania. Sponsored by the two different pharmaceutical companies, and using slightly different dose regimes, they show consistent results. Both drugs appear effective improving mania. The improvement was slightly faster and slightly greater with olanzapine (average doses 17.4 and 14.7 mg/day) than with valproate (1401 and 2115 mg/day). Insomnia, overactivity, and flight of ideas improved significantly more with olanzapine in one study. Interestingly, this superiority of olanzapine over valproate was seen only in the non-psychotic group of manic patients. The drugs seemed equally effective in psychotic mania, as they did in mixed mania.

Apart from speed of action, which is greater with antipsychotics, there are differences in side effects. Olanzapine produced more somnolence (39% vs. 21% and 47% vs. 29%), dry mouth (34% vs. 6%), running nose (14% and 3%), edema (14% and 0%), increased appetite (12% vs. 2%), and weight gain, whereas valproate produced more gastrointestinal disturbance with nausea (29% vs. 10%). Other side effects of valproate and olanzapine occur but are too rare to have been detected in these trials.

Carbamazepine or Valproate in Mania

The first parallel-group placebo-controlled trial of carbamazepine was recently published; the earlier widespread use of carbamazepine was based on evidence from crossover studies and other trial designs. Valproate is effective in a proportion of manic patients including non-responders to antipsychotic drugs and lithium. Patients who respond to valproate do not necessarily respond to carbamazepine and vice versa. In the first large parallel-group placebo-controlled study (which included only patients who were unresponsive to or intolerant of lithium), 59% of patients on valproate improved compared to only 16% of those on placebo. Most of the improvement occurred within 1 to 4 days of achieving therapeutic levels. A second and larger study comparing divalproex to lithium or placebo in a 3-week parallel-group double-blind study is shown in Table 5. Half of the patients had been unresponsive to lithium previously. Valproate was as effective in rapid-cycling mania as in other manic patients, and equally effective in the patients previously judged responders or non-responders to lithium. However, few patients in the study returned to normal functioning within 3 weeks.