CBT Plus MAOI More Effective for Social Anxiety than Either Alone
SSRIs are the mainstay of medication treatment for social anxiety disorder (SAD), while cognitive behavior therapy (CBT) is at least as effective. Oddly, combined SSRI and CBT treatment has not clearly outperformed each treatment alone in controlled trials. But a new study implies that MAOIs combined with CBT may be particularly effective.
In this study, 128 patients with SAD were randomized to one of four conditions: combined cognitive behavior group therapy (CBGT) and the MAOI phenelzine (Nardil), phenelzine alone, CBGT alone, or placebo. CBGT was given in weekly 2.5-hour sessions to groups of four to six patients for 24 weeks. Phenelzine was started at a dose of 15 mg/day and was gradually increased to a potential maximum dose of 90 mg/ day, with the final average dose ending up at about 65 mg/day. Patients didn’t take any other psychotropic medications, except chloral hydrate or zolpidem (Ambien) as needed for sleep.
Patients were assessed at six, 12 and 24 weeks with well-known anxiety rating scales. After 24 weeks, the greatest improvement in Clinical Global Improvement (CGI) response rates was seen in the combined treatment group—78.1% response rate (25 of 32 participants). Combined treatment was statistically superior to the phenelzine only group (48.6% response rate), the CBGT group (52.9%), and the placebo group (33.3%) (Blanco C et al., Arch Gen Psychiatry 2010;67(3):286–295).
TCPR’s Take: While the results were statistically significant, there were fairly high drop out rates in the various treatment arms, and the total number of participants was rather small. But the results are suggestive and imply that combining MAOI treatment with group cognitive therapy is an effective—if rarely feasible—option.
Seroquel For Bipolar Depression: The BOLDest and Best?
In 2005 and 2006, two trials showed a substantial advantage for quetiapine (Seroquel) over placebo for bipolar depression. Based on these trials, which were known by the acronym BOLDER I and II, quetiapine received an FDA indication for the condition.
AstraZeneca recently published an additional two trials, which compared the drug to lithium and paroxetine (Paxil) in the treatment of bipolar depression. The monikers this time: EMBOLDEN I, which compared quetiapine 300 mg (n=265 patients), 600 mg (n=268), lithium 600 to 1,800 mg (n=136), or placebo (n=133); and EMBOLDEN II, which compared quetiapine 300 mg (n=229), 600 mg (n=232), paroxetine 20 mg (n=118), or placebo (n=121). Both trials were eight weeks, randomized and double-blind. They used the Montgomery-Asberg Depression Rating Scale as their primary outcome measure, along with several secondary measures.
In the two trials, both quetiapine doses (300 mg and 600 mg) were superior to placebo on the MADRS and most other secondary measures, whereas lithium was no better than placebo on any measure, and paroxetine was only better than placebo on one outcome (anxiety). Across the two studies, patients on quetiapine gained about three to four more pounds than patients on placebo. Patients on paroxetine lost, on average, less than one pound relative to patients on placebo.
Dropout due to serious events was slightly higher (but not statistically significant) on quetiapine relative to lithium, but higher on paroxetine than quetiapine. Paroxetine had a higher rate of mania during treatment (9%) than either 300 mg (2%) or 600 mg (4%) of quetiapine, though these rates were not significantly different. Sleepiness, sedation, and dry mouth were the most common side effects on quetiapine (McElroy SL et al., J Clin Psychiatry 2010;71:163–174; Young AH et al., J Clin Psychiatry 2010;71: 150–162).
TCPR’s Take: These studies solidify quetiapine’s position as the medication with the largest evidence base for bipolar depression. However, we can’t resist pointing out some methodological issues. About 35% of patients on lithium did not achieve the targeted serum level of 0.6 to 1.2 mEq/L, which could have disadvantaged lithium. The side effect profiles of the medications in this study are fairly distinct, so it is possible that the raters could distinguish which patients were taking which medications, thus compromising the double-blind. Finally, rating scale differences in favor of quetiapine over lithium and paroxetine were small and generally not statistically significant. Nonetheless, quetiapine clearly appears effective in bipolar depression, with its main disadvantage being its side effects of sedation and weight gain.
Accurate Prediction of Psychosis? Maybe
The early detection of schizophrenia is a hot topic in psychiatry. If we could detect schizophrenia during the “prodromal” phase, before overtly psychotic symptoms became problematic, perhaps early intervention could prevent or delay the onset of full-blown schizophrenia. But most efforts to date have been disappointing, with high rates of false positives, people who were predicted to develop a psychotic disorder but did not actually become psychotic. For example, in one recent study only 35% of people predicted to develop psychosis actually became psychotic over a 2.5 year followup (Canon TD et al., Arch Gen Psychiatry 2008;65:28–37).
But a freshly published study may provide cause for optimism. Researchers included 245 participants, who were referred by various mental health providers or self-referred due to concerns regarding warning signs of psychosis, such as declining functioning, poor concentration, and suspiciousness. Participants were assessed with two types of ratings used in prior research: 1. ultra-high-risk criteria (UHR) and 2. cognitive disturbances (COGDIS). Follow-up occurred at nine and 18 months.
Of participants who scored positive on both UHR and COGDIS, only 24% developed psychosis at follow-up, providing an atrocious false positive rate of 76%. However, using a complex statistical model that combined features of both UHR and COGDIS, 83% of those predicted to become psychotic did so, and 87% of those predicted not to become psychotic were not psychotic at follow-up. The predictive model included positive symptoms, bizarre thinking, sleep disturbances, schizotypal disorder, functioning in the past year, and years of education (Ruhrmann S et al., Arch Gen Psychiatry 2010;67:241–251).
TCPR’s Take: This seems like a major improvement in accuracy, but there is one major caveat: The predictive model was generated retrospectively to best fit the data; it was not generated before the study began. To produce reliably credible results, the same predictive model would need to be used on a different set of participants and generate similarly strong predictive accuracy. For more discussion of early detection of psychosis, please see the December 2009 TCPR.