After 6-weeks, monotherapy lurasidone patients at a dose range of 20–120 mg/day performed significantly better and had greater reduction in functional impairment, as indicated by the SDS total score, compared to those on placebo. Similar change from baseline to 6-weeks was demonstrated across each of the SDS domain scores (work/school, family, and social life), indicating that all three domains contributed equally to the SDS total score improvement. Further, patients receiving all doses of lurasidone treatment also reported significantly fewer days lost and days unproductive due to symptoms in comparison to placebo. These domain-specific findings demonstrate that at various dosage levels, lurasidone effectively improved functional impairment in patients with bipolar disorder in all areas of impairment assessed.
Previous research has demonstrated that lurasidone significantly improved symptoms of bipolar depression and functional impairment (Loebel et al. 2014a, b). The current analysis extends these findings by assessing if improvement in functional impairment (SDS) due to lurasidone treatment was independent of improvement in bipolar depression symptoms (MADRS). Path analysis revealed that improvement in SDS total scores was largely but not completely explained by improvement in MADRS. The remaining statistically significant and medium-strength direct effect (β = −0.17), after accounting for the indirect effect of MADRS change on SDS total score change (57 % of the total effect), revealed that treatment had an independent effect on improvement in functional impairment.
These findings are supported by a recent post hoc analysis of the original 485 patients included in the lurasidone efficacy analyses reported by Loebel and colleagues (Loebel et al. 2014b). In this post hoc analysis, Loebel and colleagues estimated rates of recovery, defined by combined symptomatic remission (MADRS ≤12) and functional remission (all SDS domain scores ≤3) sustained for at least 3 months in the 6-month continuation study, in patients treated with lurasidone monotherapy (Loebel et al. 2015). The proportion of lurasidone-treated patients attaining symptomatic remission (defined as a MADRS total score ≤12) at week 6 was significantly higher (40.9 %) compared to placebo (24.7 %, p < 0.01) (Loebel et al. 2015). Similarly, the proportion of lurasidone-treated patients achieving functional remission at week 6 was significantly higher (48.4 %) compared to placebo (31.5 %, p < 0.01). However, as the current analysis utilized path analysis to assess for direct and indirect effects of treatment, it is impossible to rule out potential pseudo-specific effects on the patient’s report of functioning improvement.
The clinical importance of these findings is supported by a body of research that has documented the value of assessing functional impairment as a disability endpoint in this population. Multiple investigations have demonstrated that patients with bipolar depression are significantly more likely to report impairment in functional areas valued by patients, including relationships with family and friends, functioning at work and school, and cognitive impairment (Keck 2004; Rosa et al. 2010; Altshuler et al. 2006; Tohen et al. 2000; Simon et al. 2007; Henry et al. 2013; Depp et al. 2012; Gutierrez–Rojas et al. 2011; Calabrese et al. 2004). Further, even patients in remission from depressive symptoms may show continued, impaired psychosocial functioning (Rosa et al. 2010; Greer et al. 2010), demonstrating the need to assess functional impairment even in the absence of continued depressive symptoms. Thus, patient-centered assessment tools such as the SDS are valuable for measuring change in functional and disability outcomes important to patients with bipolar disorder that may not be captured using clinician-completed symptom assessments such as the MADRS. Our findings support the need for independent assessment patient functional improvement as an endpoint when assessing the efficacy of antipsychotic or antidepressant treatment. However, it should be noted that although the SDS was developed for the assessment of functional impairment in clinical trials of depression, anxiety, and bipolar disorders, the SDS is often found to be moderately correlated with measures of depression. Indeed, in the present study, the SDS Total score was found to be moderately correlated with the MADRS (r = 0.37). Thus, the findings of the present study are preliminary, as no clinician- or performance-based assessment of functional impairment was available to use to confirm the patient’s report of improvement in functioning specifically.
Although treatment remained a statistically significant predictor of improvement in functional impairment, changes in bipolar depression symptoms accounted for the majority of the effect on change in functional impairment. As has been elaborated by the FDA PRO guidance document (Food and Drug Administration 2009), a well-validated PRO should be able to measure the effects of a treatment on “how a patient feels or functions” both through the direct and mediated effect on symptom improvements. The findings of this analysis are consistent with previous work that has demonstrated that mood symptoms (such as depression) may actually be independent predictors of functioning in patients with bipolar depression (Burdick et al. 2010) or major depressive disorder (Sheehan et al. 2011; Wise et al. 2008). For example, Sheehan and colleagues (Sheehan et al. 2011), using the SDS to assess functional outcomes, conducted multivariate lead-lag regression analyses to demonstrate that changes in mood and depressive symptoms at weeks 6 and 7 (during an episode or in a treatment trial) were significantly correlated with changes in the SDS total score at week 8, indicating that changes in mood symptoms associated with treatment significantly predicted change in functional outcomes. Using a causal modeling approach in the current study, our findings supported those of Sheehan and colleagues, as improvement in symptoms of bipolar depression, in turn, resulted in improvement in functional impairment.
The findings from the mediation analysis have particular clinical importance as it relates to the treatment of individuals with bipolar depression. Specifically, it is possible to reduce disability and functional impairment, in conjunction with symptomatic remission with the same treatment. Indeed, Simon and colleagues (Simon et al. 2007) conducted a secondary analysis of a 12-month randomized trial of a care management and psychoeducational intervention for bipolar disorder. These researchers found that within-person improvement in depression severity due to treatment was associated with clinically significant improvement in impairment and disability. While some researchers advocate for interventions that specifically target improvement in functional outcomes (Rosa et al. 2010), our findings provide preliminary evidence for an efficient treatment that has been demonstrated to improve function, both mediating through and independently from a reduction in depressive symptoms. However, further research using a causal study design is needed to confirm these findings.
Several study limitations should be noted. First, this study was a post hoc analysis of a short-term acute trial. A longitudinal study is needed to confirm that patients maintained these improvements over a longer period of time, and to determine if the mediating effect of depressive symptom severity improvement on reduced functional impairment persists. Second, functional impairment was assessed based on patient-reported data of patients that had not yet achieved remission status, and not through direct observation of patient behavior and functioning. Similar to many registration clinical trials facing limited resource issues, functional impairment was measured by the SDS as one of the secondary endpoints, and our clinical trial did not employ additional informant assessment such as clinician-reported measures of functioning or performance-based assessments to validate the self-reported functioning outcomes. Along the same lines, no neuropsychological measures were administered to further understand the psychological functioning of the patient population at baseline or study completion. Thus, our analyses and the conclusions that can be drawn in our study are exploratory in nature and need further confirmation from future investigations. Finally, in regards to the path analysis, our sample size was limited and our tested models should be re-estimated in a larger population, whenever possible in future studies. While mediation analyses allow for the interpretation of a causal association between constructs, these findings are preliminary and further analysis using a causal study design will help to substantiate the current findings that changes in bipolar depression were causally associated with changes in functional impairment.