This is the first double blind randomized trial to compare divalproex with risperidone for the treatment of manic or mixed episodes in PBD. Our central finding supported our hypothesis that patients on risperidone improved more rapidly when compared to those receiving divalproex. This finding favors the choice of risperidone in acute mania for a more speedy response. Also, contrary to our prediction, we found that risperidone showed significantly greater baseline-to-individual endpoint reduction compared to divalproex on the YMRS and other outcome measures. In addition, the YMRS response rate and the remission rate were also significantly greater with risperidone compared to divalproex. While there were no significant differences between groups in safety, greater number of patients continued to take risperidone for the full 6 week duration of the trial, compared to those receiving divalproex. Drop out rate was 24% in risperidone group and 48% in divalproex group, with increased irritability being the most common reason for drop out in the latter. Higher retention in risperidone group may also be due to reasons such as rapid and greater response in this group, although the current study cannot offer a definitive explanation for the lower rate of subject attrition in the risperidone group.
The onset of response is vital in an acute manic or mixed phase of PBD, as risks for impulsive, destructive and suicidal behavior are high during these episodes. Our results showing a more rapid reduction in symptoms with risperidone, as measured by the YMRS, when compared with divalproex, are consistent with a previous study that compared quetiapine with divalproex for treatment of PBD (10
) and in adult studies of risperidone monotherapy (40
). Another study in adult bipolar disorder that compared olanzapine, a SGA similar to risperidone and divalproex also indicated that olanzapine had faster onset of action relative to divalproex (41
). Further, a retrospective chart review comparing risperidone with divalproex showed a faster decrease in symptoms with risperidone based on the CGI-Improvement Scale (42
). Thus, our results support previous findings indicating that manic symptoms respond more rapidly to an SGA than a mood stabilizer.
Risperidone showed greater response than divalproex in treating symptoms of mania in PBD. The results on response rate with risperidone are similar to those reported in other studies of SGAs including studies of risperidone (25
), olanzapine (26
), aripiprazole (9
), quetiapine (10
) and ziprasidone (43
). The 21.4 point drop in YMRS score with risperidone is comparable to the 23 point drop with quetiapine reported by DelBello et al. (10
) using a similar design, and the 18.5 point drop reported by the DBPC trial of risperidone with similar doses to those of the present study (25
). Similarly, an adult study that compared risperidone to placebo showed a reduction of 9.68 points among completers and 8.29 points among the group with LOCF (40
). However, the drop in response to divalproex was much lower at 13 YMRS points in this study when compared to 19 points by DelBello et al (10
). While the decrease in symptoms on YMRS with divalproex was much lower than with risperidone in our study, we found a greater response with divalproex (13 points) than was reported by Wagner et al.(8.8 points) (16
), which did not differ from improvement of those on placebo. It is important to recall that our study did not have a placebo control arm. Analytic issues may explain the difference in the findings of these two studies. Wagner et al. (44
), in their intent-to-treat analysis of efficacy, included all subjects who had at least one dose of divalproex, using last observation carried forward (LOCF) methods, whereas the present study excluded subjects who did not complete at least one weeks of post-baseline assessments and used mixed models (with all points of measurement) and repeated-measures analysis of scores between baseline and the study endpoint. In the Wagner et al. study (44
), nearly one fourth of the subjects taking divalproex had discontinued divalproex before the end of the study. This could have made an effect more difficult to detect. In a Cochrane review of divalproex, using the outcome ‘failure to respond by the end of the trial,’ divalproex was found to be efficacious in adult studies relative to placebo (Relative risk ratio i.e., RR at .62; 45;46) and was comparable to lithium in the pediatric study (RR at 1.11; 11) of acute manic and hypomanic episodes. However, another important observation is that divalproex extended release (ER) form did not yield statistically significant improvement in mania symptoms in adult patients in a recent multi-site DBPC trial reported by Hirschfeld et al (47
). It may be that the lower effect size in this study as well as that by Wagner et al (44
) may be due to the divalproex ER preparation for acute mania, as opposed to the immediate release form used in this study.
The more rapid rate of improvement with risperidone is clinically important to help patients quickly return to functioning at an optimal level. Even with a sample size of 65 subjects, we were able to show a robust and significantly greater decrease in YMRS symptoms with risperidone relative to divalproex. It is important to note that response with the low dose used in the DBPC trial of risperidone (25
) and the mean dose used in our study (1.44±0.56mg) are comparable to the response with the high dose used in the DBPC trial (25
). This suggests that there is no added advantage of using higher doses than 2.0 or 2.5 mg of risperidone in children and adolescents used in these studies. While the current sample is not sufficiently large to yield definitive results, our findings provide new information about the size of drug response that can be used to estimate the sample that will be required to better clarify the differences between these drugs.
Concerning psychosis, DelBello et al.(10
) reported no difference in outcome when comparing quetiapine and divalproex on positive and negative symptom scores (PANSS; 48). Consistent with these results, we found that risperidone and divalproex were equally effective in reducing psychosis based on the BPRS-C scale change scores. Given the broad range of symptoms assessed with the BPRS-C, a more narrow measure that captures specific symptoms of psychosis in a larger sample size may better answer this question.
While it is important to sort out the effects of medications on aggression distinct from improvement in mania, previous studies demonstrate a steady reduction in aggression, impulsivity and irritability with divalproex (10
) and risperidone (50
). Using the OAS, we found a similar pattern of reduction in aggression and irritability with both medications, and no between treatment group absolute differences.
While there were no significant group differences in adverse effects, these results may be limited by the size of the sample and non-uniform occurrence of such events. Given that more patients on risperidone remained in the study compared with divalproex, it appears that risperidone may have been better tolerated, and this may be relevant in terms of patient adherence. Prolactin was increased more than three fold in risperidone group with no clinical adverse events such as breast enlargement or lactation and these results are similar to those reported by Biederman and colleagues in two open ended studies of risperidone (26
). Prolactin elevation results from the antagonism of D2 receptors on the lactotrophs of the anterior pituitary, resulting in a disinhibition of prolactin release. Agents with the highest binding affinities for D2 (e.g., risperidone) result in the most substantial increases in prolactin levels (51
). Treatment trials of risperidone in children and adolescents show acute elevations in prolactin, which peak and then regress to lower but still elevated steady-state levels after approximately 6–8 weeks of therapy. Significant elevations above 60 ng/mL are not uncommon (52
). These, as well as long-term, studies reported by the manufacturer suggest that while prolactin elevation is significant and common, it does not always result in significant adverse-effects (e.g., galactorrhea, hirsutism, menstrual disturbances, decreased libido, etc) (53
While neither group had patients that had significant weight gain, these results must be interpreted with caution due to small sample size in this study and in the light of other studies that reported greater weight gain, both with SGA and divalproex. In addition, lower weight gain, especially in risperidone group could be partially explained by the fact that this is a short-term out patient study requiring lower mean dose of medication in addition to lower mean baseline weight relative to divalproex group (although not significantly different). Among SGAs, weight gain was around 2 kg with 3 weeks of risperidone (54
), 3.7 kg with 3 weeks of olanzapine (55
), 0.55 kg with 4 weeks of aripiprazole (56
), 1.7 kg with 3 weeks of quetiapine (57
) and 1.0 kg with 3 weeks and 2.8 kg with 27 weeks of ziprasidone (58
). In light of these findings, it is important to monitor for weight gain, glucose, and lipid parameters on regular basis (59
). With regards to EPS, 6.3% of the subjects in the current study showed symptoms which is comparable to 8% in another study using a low dose of 0.5–2.5 mg of risperidone (54
) and 12.2% on 10 mg dose of aripiprazole. Larger doses of risperidone (3–6 mg; 54) and aripiprazole (30 mg; 56) lead to higher incidence of EPS at 25% and 27.3% respectively. In contrast, quetiapine showed lowest incidence of EPS in PBD (10
This study has limitations. Despite a sample size sufficient to detect drug treatment differences based on the YMRS-derived effect sizes, the sample size was not adequate to detect differential group differences on other indices, such as the parent-report CMRS-P or the clinician-rated aggression and psychosis measures. However, the parent ratings on the CMRS-P add the value of multiple informants, which reduce but do not eliminate the potential for bias. This was a “double-dummy” design, not a placebo-controlled trial comparing a single drug with placebo which may be necessary to definitively establish efficacy. Also, this is a short trial and extended use of either medication may have yielded definitive results. It cannot be generalized to inpatient population as it is an out patient trial with dose adjustments limited to the first visit. Finally and possibly related to the small sample size, differences by active drug in response and remission may be less definitive, with greater improvement suggested for those taking risperidone. As for the clinical implications, findings from this current study indicate that, second generation antipsychotic, i.e., risperidone may be a first option due to rapid response and greater reduction manic symptoms, when compared to an anti-epileptic, divalproex. In addition, with regards to tolerability, risperidone may be superior to divalproex, with no irritability or lesser gastro intestinal side effects with risperidone. Given that we found increased prolactin with risperidone, this may warrant close follow up for persistent high levels that could lead to emergence of adverse events such as sexual dysfunction, breast enlargement or lactation.