There is a disappointing dearth of studies examining second-step medication treatment strategies in individuals with mood or anxiety disorders who are not responsive to initial evidence-based medication treatments. Those few studies that do exist have been largely focused on depressive illness. Virtually no large-scale second-step treatment studies have been conducted for any of the most prevalent anxiety disorders (panic disorder, generalized anxiety disorder, social anxiety disorder, and posttraumatic stress disorder).
In this issue of the
Journal, Pollack et al. (
1) provide important guidance for clinicians treating patients with social anxiety disorder, where nonresponse rates are a bit higher than in depression, panic, and generalized anxiety disorder (
2). The diagnosis of social anxiety disorder evolved from an initial 1960s focus on discrete, performance-related anxiety (termed “social phobia”) to a later focus on more widespread, generalized social anxiety and avoidance (
3). An often unappreciated fact is that such individuals are not only unable to work but unable to establish long-term intimate relationships (
4). This latter issue often leads many of these patients to the offices of psychotherapeutically oriented psychiatrists, and so this study should be of interest to them as well as to psychopharmacologists.
In this landmark study, the investigative team, across three geographically separate sites (Boston, San Diego, and Toronto), recruited 397 patients with DSM-IV diagnosed generalized social anxiety disorder and a history of at most two unsuccessful pharmacologic treatment trials. Participants were treated in an open-label fashion with the selective serotonin reuptake inhibitor (SSRI) sertraline at doses of at least 50 mg (maximum, 200 mg) for 10 weeks. This controlled for the most common reason for medication nonresponse: insufficient dosage or inadequate duration of treatment. The remission rate after the open-label SSRI treatment was very low (13%), and only 32% achieved a “response” (Liebowitz Social Anxiety Scale score, <50). Following this phase, 181 patients with an inadequate response to treatment (Liebowitz Social Anxiety Scale score of 50 or greater, indicating at best a partial response to the medication) were randomly assigned to one of three strategies commonly employed after nonresponse to a standard SSRI: augmentation with a benzodiazepine, in this case clonazepam; switch to a different antidepressant, in this case the serotonin-norepinephrine reuptake inhibitor (SNRI) venlafaxine; or continued treatment with the SSRI sertraline for another 3 months. As the first large-scale next-step medication treatment study in an anxiety disorder patient group, this study not only has clinical importance but it provides results that should guide the next phase of research in this area.
Significant differences were observed in both Liebowitz Social Anxiety Scale symptom levels and Sheehan Disability Scale scores between patients receiving augmentation with clonazepam and patients continuing with sertraline alone. The venlafaxine-treated patients had an intermediate outcome, with their symptom and disability scores not significantly different from either of the other two groups, although outcomes were closer to the sertraline group. The overall patient response rates are instructive: 27% with clonazepam, 19% with venlafaxine, and 17% with sertraline plus placebo. Merely waiting an additional 10 weeks and continuing sertraline produced an additional 17% of responders, much greater than the 10% margin of effect between clonazepam and sertraline plus placebo. While a number of studies in anxiety disorder patients have shown that longer duration of treatment with SSRIs will yield additional responders (
5,
6), this fact is often overlooked by clinicians pressured to provide more or different medications to anxious patients who are frustrated with their lack of short-term response. The limited difference between the two antidepressants is also noteworthy and suggests that switching to a different antidepressant class is less effective than augmenting with a benzodiazepine. While some may argue that higher doses of venlafaxine could have achieved greater effect (the mean doses were 168 mg), this notion is thought to be more relevant for the treatment of depression (many anxiety disorder experts feel that higher doses of SNRIs are anxiogenic because of greater noradrenergic effects at these levels). The clonazepam group, compared with the other two groups, had numerically fewer adverse events, except for somnolence, in multiple categories. Hence, this strategy was both more effective and probably more tolerable. The incidence of insomnia was markedly lower in the clonazepam group (13% compared with 42% and 36%), which is instructive in light of recent evidence that insomnia seems to predict poor response to CBT in social anxiety (
7).
The Pollack et al. study is also the very first benzodiazepine augmentation study in anxiety disorder patients who did not respond to an initial antidepressant. As recently as 1 year ago, a review of treatment strategies in social anxiety disorder highlighted confusion about the potential role of benzodiazepines for social anxiety disorder because of the lack of published clinical trial data (
8). This absence of data is astonishing, given that combined benzodiazepine and SSRI pharmacotherapy is the medication strategy most commonly employed by psychiatrists treating social anxiety disorder (
9). In the absence of evidence, psychiatrists will often use “practice-based evidence” and do what is easier, more familiarly effective in their hands, and associated with fewer risks of adverse events and toxicity. That benzodiazepine augmentation wins out over antidepressant switch or continuation is important, as it finally provides an evidence base to support this common practice, even as it both confirms and denies clinical lore about the utility of benzodiazepines for anxiety (i.e., while this strategy produced statistically greater response and symptom improvement, it did not produce greater rates of remission or a “cure”).
Benzodiazepine pharmacotherapy has had a checkered history for over half a century. After initial widespread use in generic anxiety states because of their greater margin of safety compared with older sedative hypnotics, prominent publicity about their adverse effects led them to fall out of favor in the 1970s. With studies showing that alprazolam monotherapy was surprisingly effective in patients with disabling panic disorder in the 1980s, they once again gained ascendency in use. By the mid 1990s, although SSRIs had replaced benzodiazepines as the recommended first-line monotherapy for anxiety, they continued to be the most widely prescribed class of pharmaco-therapeutic agents, most often as adjunctive agents prescribed with SSRIs (
9). Clinicians skeptical of the efficacy of these agents often suggest that they merely treat nonspecific symptoms of anxiety such as insomnia and restlessness—“taking the edge off” of symptomatic distress without affecting the core syndrome. The Pollack et al. study findings are not consistent with this bias, since response on the Liebowitz Social Anxiety Scale required significant change in multiple aspects of the core social anxiety disorder syndrome. There is now general appreciation that, with careful clinical evaluation to rule out pre-existing addition issues and careful monitoring for cognitive and psychomotor effects, especially in older individuals, benzodiazepines are safe and highly effective, as this study demonstrates. While this study cannot inform us about the knotty issue of benzodiazepine dependence and how many patients would have great difficulty tapering off of these agents, long-term therapy in other anxious patient groups has not been associated with dose escalation and tolerance (
10) (i.e., they can likely be taken safely for years in appropriately selected patients).
In their discussion, Pollack et al. refer to the possible benefit of additional “psychosocial therapeutic strategies.” Combining medication with psychotherapy in social anxiety disorder produces better outcomes than either treatment alone (
11), suggesting that future second-step treatment studies should test how adjunctive use of additional medication compares with adding psychotherapy. The recent finding that adding exposure and response prevention was superior to adding risperidone for individuals with obsessive-compulsive disorder who did not respond to SSRIs (
12) disproves concerns that psychotherapy is not useful for more severely ill and treatment-resistant patients. The Pollack et al. study can now inform the choice of that adjunctive medication, i.e., a benzodiazepine, which could be compared with CBT (probably the best choice since a recent study found that it was superior to psychodynamic therapy [
13]). Including a third option of combined CBT and benzodiazepine would address longstanding concerns about benzodiazepines interfering with the efficacy of CBT (
14) (however, there is no consistent evidence for this) and address the hypothesis that combined medication and psychotherapy treatment might be superior in phase 2, just as it has been in phase 1 (
10). This would likely not only advance the science but also the clinical practice of treating social anxiety disorder.