1 INTRODUCTION

It is well-established that a substantial proportion of individuals with mood disorders have objective cognitive impairment,^{1, 2} with impairment present across a range of cognitive domains, both in episode and euthymia.^3-6 Cognitive impairment is linked to poorer functional outcomes^7-9 and an increased likelihood of episode recurrence.^{10, 11} In this context, there has been increasing interest in examining the potential moderating effects that cognitive functioning has on treatment outcomes in bipolar disorder (BD) and major depressive disorder (MDD). A greater understanding of cognitive predictors of treatment response may help delineate particular individuals who require more intensive intervention to improve aspects of cognition.

A systematic review has demonstrated that certain domains of cognitive function may be associated with treatment outcomes in mood disorder populations, particularly attention/executive function, and processing speed.¹² Much of this research has focused on predictors of pharmacotherapy response. Few studies have examined this issue in relation to psychotherapy. The psychotherapy response literature in both BD and MDD reports that the most convincing evidence is for a relationship between poorer baseline verbal learning and memory performance and less improvement in mood symptomatology post-treatment.^13-16 However, not all studies have found this,^17-19 and other cognitive domains have also shown predictive utility, for example, processing speed^{16, 19, 20} and executive function.^{15, 16, 21} Literature on cognitive predictors of functional outcome is even more sparse, with little consistency around which cognitive domains are predictive.^{14, 19, 22-24} For example, Deckersbach et al.¹⁴ found that poorer performance on tasks of executive function and memory at baseline predicted poorer functional outcomes following cognitive remediation (CR) in BD. In contrast, a comparison of CR versus treatment as usual in BD showed a relationship between poorer cognitive performance, including executive function and psychomotor speed, at baseline and greater changes in psychosocial functioning at treatment-end.²⁴ Further research is therefore required to examine this issue fully.

In consideration of the paucity of research examining cognitive predictors of treatment outcome following psychotherapy, the aim of this exploratory analysis was to examine associations between baseline cognitive function and treatment outcome in individuals with mood disorders receiving 12 months of Interpersonal and Social Rhythm Therapy (IPSRT). In contrast to most of the literature, both mood and functional outcomes were examined in the current study. A measure of mood symptom burden as opposed to a cross-sectional measure of mood symptom severity was employed as the former has been identified as a strong predictor of patient functioning and distress.^{25, 26} Whilst a cross-sectional measure assesses mood symptomatology at a single time point, the Longitudinal Interval Follow-up Evaluation (LIFE)²⁷ interview assesses mood state retrospectively over a period of time (usually 6 months), yielding a number of measures including an overall score which is indicative of overall burden of mood symptoms. While the LIFE is a retrospective interview, we have recently shown a good correlation with concurrently measured mood symptoms.²⁸

Primary outcomes from the original randomised controlled trial (RCT), including change in cognitive function, are reported elsewhere.²⁹ In brief, one of the key findings was that the addition of CR to IPSRT resulted in less improvement in mood symptoms and psychosocial functioning at treatment-end. The primary goal of this analysis was to examine the association between baseline global objective cognitive function and mood symptom burden, as defined by the LIFE at treatment-end (12 months). In addition, we examined the relationship between (a) individual domains of objective cognitive function performance at baseline and mood symptom burden and functional outcome at treatment-end, and (b) subjective cognitive performance at baseline and mood symptom burden and functional outcome at treatment-end. Baseline cognitive scores were standardised according to performance of a healthy control group which also gave an easily interpreted statistic reflecting the degree of impairment of patients at baseline.

2 METHODS

3 RESULTS

Of the 68 mood disorder participants who entered the study, 58 completed treatment and follow-up testing (IPSRT = 31, IPSRT-CR = 27). Forty-eight healthy control participants were also recruited (see Table 2). The mean age of the mood disorder sample was 36.3 years (SD 12.3), and 34 were female (59%). Thirty-six (62%) individuals had a diagnosis of BD (17 BD I, 18 BD II and 1 BD NOS) and 22 (38%) had a diagnosis of MDD. Thirty-four (59%) were in full or partial remission at baseline, and 24 (41%) were in episode. The mean QIDS-C score was 7.66 ± 5.5 (range: 0–20).

The mean (SD) z-scores for each domain representing an effect size difference between the mood disorder participants were: global cognition = −0.29 ± 0.56, verbal learning and memory = −0.37 ± 0.73, visuospatial learning and memory = −0.48 ± 1.12, attention/executive function = −0.43 ± 0.69 and psychomotor speed = 0.14 ± 0.94. Values for individual variables are provided in Table S1. The mean (SD) z-score for the COBRA was −2.10 ± 1.59.

4 DISCUSSION

The aim of the current analysis was to examine associations between baseline cognitive function and change in mood symptom burden and functioning following a 12-month course of IPSRT. We found that baseline global objective cognitive function was not predictive of change in mood symptom burden post-treatment. In analyses of individual cognitive domains, poorer baseline attention/executive function was associated with less change in total mood symptom burden, and burden of depression symptoms, following psychotherapy. Slower baseline psychomotor speed was also associated with less change in burden of mania symptoms by treatment-end.

The participants with mood disorders were recruited on discharge from Specialist Mental Health Services. In New Zealand, only people with severe mental illness are seen by Specialist Mental Health Services, with most others being treated in primary care. Specialist Mental Health Services include inpatient care and community mental health care. Patients cared for by community mental health teams are admitted to the teams from inpatient care or direct from primary care. Patients are discharged from community mental health teams after what is usually a short (<3 months) period of stabilisation, but not necessarily after an episode has resolved. In this study, approximately 40% of patients were still in-episode. This is therefore a study of people with severe or complicated illness, at a stage of partial treatment following an acute episode of mood disturbance.

The finding of an association between poorer baseline attention/executive performance and less change in mood symptom burden at treatment-end, particularly depression symptoms, is consistent with prior research examining cognitive predictors of treatment response.¹² There are several possible explanations for such a finding. First, it is possible that poorer attention/executive function performance serves as a proxy for dysfunction within the fronto-limbic circuits. Attention/executive function is sub-served by areas within the prefrontal cortex, with aberrant neural activity having been found in these brain regions in mood disorder populations.⁴⁷ Attenuated prefrontal control over hyperactive limbic and subcortical structures (e.g. amygdala) may lead to the negative affective biases and perseverative thinking patterns that are common in mood disorders, with such cognitive processes believed to contribute to the development and maintenance of mood symptomatology.^{48, 49} Thus, poorer attention/executive function performance may highlight key pathological processes that serve to perpetuate mood symptoms and impair response to treatment.

Second, it is possible that impairment in attention/executive function might impinge on an individual's ability to effectively engage in psychotherapy and thus, derive as much benefit. Individuals in the current study underwent 12 months of psychotherapy with or without CR, which will have required a level of cognitive function necessary for the assimilation and implementation of core concepts and strategies. Impairments in attention/executive function may make it particularly difficult for individuals to engage in therapy, for example, due to difficulty attending to content discussed or lack of cognitive flexibility limiting the ability to transfer strategies learnt in therapy to real life.

There was no association between baseline subjective cognitive performance and change in mood symptomatology following therapy. This was somewhat surprising as it has been shown that subjective cognitive impairment may be related to factors such as depressive symptom severity and increased self-referential processing, albeit cross-sectionally.^50-52 To our knowledge, no previous studies have examined the predictive value of subjective cognitive complaints and treatment outcomes in mood disorder populations. Thus, further research is required to examine this relationship more fully.

No significant relationships were found between baseline cognitive performance (both objective and subjective) and changes in functional outcome as measured by the FAST and SAS. This is despite the well-substantiated link between cognitive function and functional disability in mood disorders.^{7-9, 53, 54} We are aware of five studies that have examined the relationship between baseline cognitive function and functional outcome following psychotherapy, and there has been little consistency regarding which cognitive domains are predictive or the directionality of the associations found.^{14, 19, 22-24} This may relate to the difficulties in defining and measuring functional outcome.⁵⁵ Based on current available evidence, it is apparent that the relationship between baseline cognitive function and psychosocial functioning following treatment remains unclear.

Across all three significant models, higher mood symptom severity at baseline was associated with greater improvement in mood symptom burden at treatment-end. This is not surprising given there would have been greater scope for improvement in this regard, albeit such an association may not have been found in a more clinically ill sample. For the association between baseline attention/executive function performance and change in LIFE depression scores, those in the IPSRT group showed greater improvement in depressive symptoms at treatment-end compared with the IPSRT-CR group. As noted above, the original outcome paper showed that those receiving adjunctive CR had less improvement in mood symptoms and psychosocial functioning following treatment.²⁹

Psychomotor speed was also associated with treatment outcome, whereby slower psychomotor speed at baseline was associated with less improvement in LIFE mania symptoms post-treatment. This was moderated by the presence of a concurrent substance use disorder, which was associated with a greater reduction in mania symptoms at treatment-end. These findings should be interpreted cautiously. Baseline mania scores were low, giving little scope for change, and there was also very little difference between controls and mood-disordered patients in psychomotor speed. In this context, the clinical significance is unclear and as we note in the limitations section, there is the risk of Type 1 error in this exploratory analysis. It is possible that improved management of the substance use disorder via therapy provided greater scope for improving mood symptoms, particularly mania, as they commonly co-occur. Alternatively, participants may have learnt alternative strategies during therapy to manage their mania symptoms, leading to a reduction in substance use and overall symptomatic improvement. A systematic review suggests equal, if not better, outcomes for individuals with comorbid BD and substance use disorder receiving psychotherapy,⁵⁶ while in our previous studies using IPSRT, there was no evidence that current substance use disorder impaired response.⁵⁷

The current analysis has several strengths. First, few studies have examined baseline cognitive predictors of mood and general functioning with psychotherapy across the mood disorder spectrum. Second, the current analysis examined change in mood symptom burden (both syndromal and subsyndromal) over the 12-month treatment period, which may be more sensitive to change than the more commonly used outcome of relapse and/or remission. Moreover, research suggests that patients with mood disorders can experience significant periods in subsyndromal mood states,^58-60 with ongoing mood symptom burden having been identified as an important determinant of functioning and a significant contributor to patient distress.^{25, 26} To this end, employing a longitudinal measure of mood disturbance may be more clinically meaningful for patients and may also provide a more accurate depiction of an individual's symptoms and course than a cross-sectional measure. Finally, predictors of treatment outcome were examined in a diverse group of individuals, particularly regarding clinical characteristics, thereby increasing the generalisability of study findings.

The current analysis also has several limitations. First, the sample was heterogeneous regarding multiple clinical characteristics including mood disorder diagnosis, medications and mood state at baseline. While this heterogeneity may be positive for a pragmatic clinical trial generating generalisable clinical results, it reduces the likelihood of providing clear data on predictors, particularly with a relatively small sample. Second, sample size is clearly a limitation. To mitigate this, the data were analysed together for both treatment arms, using treatment arm as a fixed factor in analysis. Using the whole sample increased power to examine factors operating across both groups but did not of course compensate for low power in examining predictors specific to those receiving CR. Thirdly, no correction was made for multiple comparisons. Whilst we attempted to mitigate this issue by using a priori-defined domains for statistical analysis to reduce the number of comparisons, the lack of correction could mean that some findings are due to chance. Finally, as the original trial was designed to be a pragmatic clinical trial with minimal inclusion/exclusion criteria, we only recruited patients who had subjectively assessed cognitive difficulties as measured by a single yes/no question. We were unable to examine the correlation between this single-item question and more in-depth measures of subjective cognitive impairment such as the COBRA as all participants were reporting cognitive difficulties. Additionally, whilst the correlation between subjective and objective cognitive difficulties is relatively small in MDD and BD,^{51, 52} this inclusion criterion may have reduced the range of cognitive difficulties in our sample, excluding patients with superior cognitive functioning and limiting the predictive power of our analyses.

5 CONCLUSIONS

This exploratory analysis found evidence that poorer baseline attention/executive function performance was associated with less change in mood symptom burden, particularly depressive symptoms, following 12 months of psychotherapy. Slower baseline psychomotor function was also associated with less change in mania symptomatology at treatment-end. The finding that poorer attention/executive function was associated with less improvement is consistent with the emerging literature on cognitive predictors of treatment response.¹² In this study, cognitive rehabilitation methods were added to psychotherapy at the beginning of treatment, partly in the expectation that this would improve cognitive functioning including attention/executive function. However, not only was CR ineffective in improving attention/executive function,²⁹ but those receiving IPSRT alone had greater improvement in mood and psychosocial functioning at treatment-end. Thus, while targeting these cognitive functions with CR in order to improve treatment outcomes may appear to be a logical conclusion from our data, the original study did not find evidence of this. The reasons for this have been more extensively discussed in our previous report.²⁹ Further large-scale studies are required to examine the role that baseline cognitive function has in determining various aspects of mood disorder recovery.

CONFLICT OF INTEREST STATEMENT

KD, CB and RP use software provided free-of-charge by Scientific Brain Training Pro for Cognitive Remediation trials. RP has received support for travel to educational meetings from Servier and Lundbeck. CB has grant support from Lundbeck, Takeda and Pfizer and has received consulting fees from Boehringer Ingelheim, Pfizer, Lundbeck.