Anxiety, irritability, and agitation as indicators of bipolar mania with depressive symptoms: a post hoc analysis of two clinical trials

Bipolar I disorder (BD-I) is an illness that comprises more than just two discrete poles; between the extremes of ‘pure’ mania and ‘pure’ depression, patients can experience ‘mixed’ states in which various manic and depressive symptoms are present at the same time. Mixed states were described by Kraepelin in the early 20th century, and have been well characterised and operationally defined in the 100 years since (Kraepelin 1921; Swann et al. 2013). The fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) defined a ‘mixed episode’ as the coexistence of the full symptomatology of a manic and a depressive episode—a restrictive, categorical approach that neglects those patients with subsyndromal depressive symptoms (Vieta and Valentí 2013; American Psychiatric Association 1994). Since the DSM-IV, studies have shown that this definition does not reflect clinical reality and that patients with bipolar disorder [as well as patients with major depressive disorder (MDD)] have a complex symptom presentation (Suppes et al. 2005; Vieta and Valentí 2013; Targum and Nierenberg 2011). Consequently, the fifth edition of the DSM (DSM-5) changed to a dimensional approach for the diagnosis of BD-I (Vieta and Valentí 2013). The DSM-5 includes a ‘with mixed features’ specifier that applies to a manic or hypomanic episode with three or more depressive symptoms (or, alternatively, to a depressive episode with three or more manic symptoms) (American Psychiatric Association 2013). In a retrospective chart review, three times as many patients with bipolar disorder were diagnosed with mixed features using the DSM-5 criteria compared with the DSM-IV text revision (DSM-IV-TR) criteria for a mixed episode (Shim et al. 2015). In an observational study, around half of patients with DSM-IV bipolar mania did not meet the criteria for DSM-5 bipolar mania, due to the additional requirement of increased activity or energy levels (Machado-Vieira et al. 2017). However, 45% of these patients now met the DSM-5 criteria for a major depressive episode with mixed features, showing the value of the mixed features specifier for making diagnoses across the continuum of mood disorders (Machado-Vieira et al. 2017).

Patients with BD-I mixed states have a worse clinical course than those with ‘pure’ manic states, since they tend to experience more episodes, and since episodes are of longer duration (Swann et al. 2013). Mania with depressive symptoms is also associated with a high rate of comorbid conditions, such as anxiety disorders and substance abuse (Swann et al. 2013). These patients are harder to treat and physicians of patients meeting the DSM-5 criteria for mania ‘with mixed features’ report greater treatment dissatisfaction than do physicians of patients without mixed features (Swann et al. 2013; Young and Eberhard 2015). Furthermore, there is a higher risk of suicidality among patients with mixed features; these patients are more likely to have made at least one suicide attempt during their lifetime, and during their current manic episode, compared with patients without mixed features (Swann et al. 2013; Young and Eberhard 2015). Thus, the presence of depressive symptoms during a manic episode is indicative of a more severe form of BD-I.

The symptoms observed in manic episodes and depressive episodes are not mutually exclusive, with anxiety, irritability, and agitation (AIA), for example, occurring in both mania and depression. Since they are not specific to either pole, AIA symptoms are not included in the DSM-5 ‘with mixed features’ specifier (Cassidy 2010; Koukopoulos et al. 2013). [Anxiety is, however, covered by the DSM-5 ‘with anxious distress’ specifier (American Psychiatric Association 2013)]. According to physicians, AIA symptoms show an increased prevalence, and greater severity, in patients with mixed features in BD-I mania compared with those without mixed features (Young and Eberhard 2015). Similarly, in an online survey, patients with mania with depressive symptoms self-reported the combination of ‘anxiety or worry’ and ‘irritability or agitation’ more frequently than those without depressive symptoms (Vieta et al. 2014). A recent prospective study from the Stanley Bipolar Network found increased irritability in bipolar patients with mixed depression compared with pure depression (Miller et al. 2016). Due to their high frequency among such patients, AIA symptoms could potentially be used as indicators (albeit non-specific indicators) to aid physicians in the identification of patients whose mania is complicated by depressive symptoms.

Using data from two pivotal, randomised, placebo-controlled asenapine trials, the aims of the present post hoc analysis were (1) to describe the phenomenology of bipolar mania in terms of AIA and depressive symptoms, (2) to evaluate if the presence of AIA and depressive symptoms can be used to predict non-remission, and (3) to evaluate the effect of asenapine, placebo, and olanzapine on remission in patients with AIA, and in patients with depressive symptoms.

This was a post hoc analysis of patient-level data from two 3-week, randomised, placebo-controlled trials (ARES 7501004 and ARES 7501005; ClinicalTrials.gov identifiers NCT00159744 and NCT00159796, respectively) of asenapine in patients with BD-I experiencing an acute DSM-IV manic or mixed episode (McIntyre et al. 2009, 2010). Olanzapine was included as active comparator.

This post hoc analysis of two randomised, controlled trials revealed that symptoms of anxiety, irritability, and agitation were present in around two-thirds-to-three-quarters of patients with BD-I mania. In addition, around a third of patients had all three AIA symptoms. While different studies have defined the presence of AIA symptoms in different ways, there is agreement in the literature that AIA symptoms are common among patients with BD-I mania (Hantouche et al. 2006; Vieta et al. 2014; Young and Eberhard 2015).

The present analysis also showed that half of patients with BD-I mania had at least three depressive symptoms, whereas less than one in seven patients had a MADRS Total score of ≥20. A previous post hoc analysis in the same dataset found that 4–34% of patients met proxy criteria for the DSM-5 ‘with mixed features’ specifier, depending on the severity cut-offs used (McIntyre et al. 2013a). In general, the incidence of depressive symptoms among bipolar disorder patients with mania is around 30–40%, though the incidence varies widely according to how the presence of depressive symptoms is defined (McElroy et al. 1992; Akiskal et al. 1998; Hantouche et al. 2006; Azorin et al. 2009; Vieta et al. 2014; Young and Eberhard 2015; McIntyre et al. 2015). In a prospective study of patients with bipolar hypomania, the likelihood of having depressive symptoms was significantly greater for women (72%) than for men (42%) (Suppes et al. 2005).

With regard to the relationship between AIA and depressive symptoms, the analysis showed that anxiety, irritability, and the AIA combination—but not agitation alone—were more common in patients with depressive symptoms at baseline. This replicates the findings of a naturalistic study, in which anxiety and irritability were more common in BD-I patients with three or more depressive symptoms, whereas agitation was not (Young and Eberhard 2015). Thus, anxiety and irritability, and the combination of all three AIA symptoms, could potentially be used as non-specific indicators, or ‘gateway symptoms’, to help psychiatrists identify bipolar mania with depressive symptoms (Eberhard and Weiller 2016). At present, the DSM-5 does not highlight or discuss AIA symptoms in the ‘with mixed features’ specifier section. Future iterations of the DSM may want to consider how to discuss the findings of this study and related studies in terms of the mixed specifier. While our findings are not definitive, given the retrospective nature of these analyses and limits of the studies themselves, they do highlight the need for further assessment of the best components to make up the mixed specifier for future DSM-5 editions.

These results support the importance of assessing carefully any patient presenting with significant anxiety, irritability, and agitation for underlying symptoms of depression, especially during mania or hypomanic episodes. While it is routine practice to administer depression scales in some settings, this is not universal. The presence of these symptoms supports that it is a time to review carefully whether they herald less visible symptomology that could inform both clinical care and address patient concerns such as self-harm. As noted in a recent article by Solé and colleagues (2017) and consistent with the findings reported here, indicators of mixity may be key to appropriate treatment choices and addressing suicide prevention effectively (Solé et al. 2017).

Of interest, a recent pilot study examined the use of actigraphy to assess prediction for different bipolar clinical states including mania, depression, and mixed (Scott et al. 2017). In this small sample with 24 h of observation, the researchers were able to predict with 79% accuracy which state a subject was experiencing (Scott et al. 2017). It is to be hoped these types of objective measures may assist us to understand, prevent, and treat mixed symptoms more efficiently.

The presence of anxiety in patients with bipolar disorder has previously been associated with a longer time to remission (Feske et al. 2000). The present analysis extended this observation—showing that patients with anxiety or a combination of all three AIA symptoms are less likely to achieve remission than those without the corresponding symptom(s). Thus, these data support the notion that AIA symptoms identify a more severe or complicated phenotype of mania that is less likely to respond to treatment.

Similarly, patients with three or more depressive symptoms were less likely to achieve remission, although this outcome did not meet the criteria for statistical significance. However, patients with a MADRS Total score of ≥20 showed similar remission rates to those below this depression threshold. The previous post hoc analysis in this same dataset also found that the remission rate (defined as a YMRS Total score of ≤12) was not influenced by the number or severity of depressive symptoms (with the exception of the most severely depressed patients, who showed a higher remission rate) (McIntyre et al. 2013a). It may be that those patients who are the most ill may also be the most responsive to the medications under study; and thus most likely to show a larger degree of improvement.

In BD-I (unlike in MDD) there has been relatively little prior investigation into whether patient characteristics (including AIA) within an episode can affect treatment response and other health outcomes (Feske et al. 2000). A recent analysis showed that the presence of AIA symptoms raises the already high suicide risk among patients with BD-I mania with depressive symptoms (Eberhard and Weiller 2016). Additionally, a subtype of bipolar depression, characterised by anxiety and irritability among other symptoms, is associated with a greater likelihood of receiving government financial support, suggesting that this pattern of symptoms is particularly malignant (Perich et al. 2016). From a clinical point of view, given the hazards associated with AIA, clinicians should routinely probe for AIA, and should systematically target these symptoms with treatment. In MDD, for example, many physicians reserve adjunctive antipsychotics for use in patients presenting with agitation or irritability, among other symptoms (McIntyre and Weiller 2015). Furthermore, the 2015 update to the Florida Best Practice Psychotherapeutic Medication Guidelines provides different algorithms for the treatment of MDD according to the presence or absence of mixed features (University of South Florida 2015); this is an unmet need for BD-I.

In the present analysis, both asenapine and olanzapine demonstrated higher remission rates (defined as a YMRS score of <12) than placebo among patients with AIA symptoms. A more stringent definition of remission was also considered (a YMRS score of <8); remission rates were low (<25%) with this definition regardless of treatment, demonstrating the difficulty in reducing symptoms to this level over such a short time period. Asenapine and olanzapine also showed consistent benefits over placebo in terms of mean change from baseline in YMRS Total score among patients with anxiety and all three AIA symptoms. In general, second-generation antipsychotics are efficacious in the treatment of mixed episodes with predominant manic symptoms, and have been described as the treatment of choice in the management of bipolar disorder mixed states (Muralidharan et al. 2013; Fagiolini et al. 2015). In mania with DSM-5 mixed features, asenapine, aripiprazole, and olanzapine have been evaluated and have shown efficacy as potential treatment options (McIntyre et al. 2013a, b; Tohen et al. 2014).

In line with considerations of best treatment choices, a range of guideline efforts for the treatment of bipolar disorder have recently been developed (Stahl et al. 2017; Fountoulakis et al. 2016; Goodwin et al. 2016) or are in revision, such as the CANMAT guidelines (Yatham et al. 2013). The importance of the results reported in this present post hoc analysis is highlighted as the best definition for mixed symptoms is sought and the understanding of the implications for treatment and course of illness is developed. It is clear from both past and more recent efforts that mixed symptoms herald a more severe course of illness, but the optimal definition is still under enquiry (Swann et al. 2013; Barbuti et al. 2017).

This present study is limited in that it is a post hoc analysis of studies that were not designed for this purpose. In particular, as these were regulatory studies, patients with current suicidality or complex comorbidities were excluded from study entry, thus limiting generalizability. In terms of study design, participants were inpatients for at least the first week of treatment, which may have selected for agitation. The results may also have been influenced by the use of benzodiazepines for agitation. Benzodiazepines were not permitted beyond Day 7 of treatment; however, 131 patients (13.6%) in the FAS received one or more doses of benzodiazepines beyond Day 7. The use of proxies for AIA and depressive symptoms was another limitation of the analysis—especially for agitation, which was based on the YMRS increased motor activity–energy item because no agitation item was available. Since irritability and agitation were based upon YMRS items, they were not independent with regard to YMRS remission. It is important to keep in mind that this proxy for depression symptoms (MADRS ≥3), and thus potential mixed symptoms, does not itself assess for the DSM-5 Mixed Specifier criteria and must be viewed as a limit of the study design. Finally, the analysis does not rule out the possibility of pseudospecificity: that AIA and depressive symptoms are discrete, overlapping phenomena.

Assessment of AIA symptoms in bipolar mania could enable physicians to identify patients with more severe depressive symptoms and who may be at increased risk of suicide, allowing for appropriate intervention. Furthermore, assessment and monitoring of AIA may help physicians to predict which patients may be harder to treat.