Cognitive Function as a Predictor of Short-Term Pharmacological Treatment Response in Major Depressive Disorder: Mediating Effect of Mentalization
Article information
Abstract
Objective
Deficits in social cognition (mentalization) and other cognitive deficits have been reported in patients with major depressive disorder (MDD) and may influence treatment response. This study examined the impact of cognitive function on treatment response of patients with MDD after 8 weeks of medication and whether the impact was mediated by mentalization.
Methods
Cognitive function (memory, attention, executive function, processing speed) and mentalization were measured in 28 patients with MDD at baseline using neuropsychological tests and self-report scales. The treatment response was defined as the rate of improvement in symptom severity and global function.
Results
Multiple regression analyses, controlling for mentalization and cognitive function, separately revealed that delayed recall was a negative predictor of functional improvement after 8 weeks of treatment, while mentalization was a positive predictor. A single mediation model using PROCESS macro showed that delayed recall and Digit Span backward indirectly affected functional improvement, mediated by mentalization. When age at onset was controlled for as a covariate, the mediating effect lost significance, and the direct effect of delayed recall on functional improvement was still significant.
Conclusion
Despite the small sample size, our results provide evidence that patients with MDD and low memory (delayed recall) at baseline may benefit more from short-term pharmacological treatments.
INTRODUCTION
Residual functional impairment that does not improve after symptom remission is associated with relapse in major depressive disorder (MDD) [1]. Therefore, the ultimate goal of treatment should be the restoration of functioning beyond symptom remission [2], and understanding the factors that critically influence treatment response, which is of paramount importance [1].
Approximately 60% of patients do not get better after their first medication and up to one-third of them do not improve with a combination of other antidepressants [3,4]. Clinical preedictors of poor treatment responses in MDD include symptom severity, suicidality, number of previous major depressive episodes (MDEs), onset before the age of 18 years, comorbid anxiety disorder (phobias and panic disorder), comorbid personality disorders, history of psychiatric hospitalization, presence of melancholic features, and failure to respond to the initial medication [5]. The identified factors are helpful in predicting treatment response; however, they are not reversible, which is a limitation preventing them from being the focus of therapeutic interventions. Predictors that can be changed through intervention have the advantage of being applied to enhance treatment response.
Cognitive function can be a predictor of treatment response because cognitive deficits are a core feature of MDD, and deficits in certain domains that persist even after symptom remission may impede recovery in patients with MDD [6]. Patients with MDE exhibit deficits in attention, memory, executive function, and processing speed [7-11]. A meta-analysis of cognitive deficits in patients with first-episode MDD [12] found that deficits in attention and executive function persisted after symptom remission, suggesting that they may be a trait of MDD. Studies have reported that cognitive deficits are associated with a poor response to treatment. For example, one study found that elderly patients with MDD and executive function deficits at baseline had poorer treatment response [13]. A recent systematic review suggested that baseline cognitive function, particularly executive function, may be useful in predicting treatment response in depression and concluded that cognitive deficits may become targets for pharmacological or psychological treatments [14].
Mentalization is related to an individual’s ability to integrate knowledge about one’s own and others’ mental states [15,16], whereas Theory of Mind (ToM) is the ability to interpret behavior by focusing on the cognitive aspects of others; mentalization involves emotional aspects and self-experience. Mentalization requires a more complex understanding of the self and others than ToM, as it not only allows individuals to guide themselves in social situations but also promotes the development of a sense of self [17]. A recent emphasis on improving mentalization in the treatment of depression suggests that mentalization impairments may contribute to poor social function in patients with MDD [18]. Patients with depression have difficulty recognizing that negative feedback in social situations is a product of others’ internal representations, and tend to accept feedback as an objective reality about themselves, which can increase depressed moods and consolidate a negative self-image [19]. A review [17] of studies that measured mentalization using the Reflective Functioning Scale found different levels of mentalization impairment among patients with depression. Those with relatively mild mentalization impairment were outpatients; in contrast, those with more severe impairment, with a Z-score of -4.43 were either inpatients with chronic course [20] or treatment-resistant patients [21]. These findings suggest that chronicity or poor treatment response is associated with more severe mentalization impairment [17]. Therefore, mentalization should be considered as a predictor of treatment response.
Mentalization is related to the interaction of various brain structures, including the medial prefrontal cortex, temporoparietal junction, temporal pole, and precuneus [22], and this neural basis suggests the possibility of interactions between mentalization and cognitive function [23]. A study reported a significant association between mentalization and executive function in patients with MDD [24]; however, the related literature is lacking. Relatively more studies have examined the relationship between social cognition (ToM and emotion perception) and cognitive function in patients with MDD, but the results have been mixed. For example, difficulties in emotion perception through prosody have been associated with poor executive function (cognitive inhibition, mental setshifting, and working memory) [25], and poor performance on ToM tasks has been associated with difficulties in working memory, logical memory, and cognitive inhibition [26]. A recent study with young adults and adolescents in MDE suggested that diminished executive function (cognitive flexibility) could lead to rigid interpretation of social cue, which resulted in lower performance on social cognition task (ToM) [27]. However, there are results of no association between social cognition and cognitive function [28,29]. Furthermore, executive function (planning) was associated with lower performance on a facial emotion perception task [27]. As for other mental disorders, a study that examined relationship between social cognition and cognitive function in patients with borderline personality disorder found that executive function and a full-scale intelligence quotient predicted social cognition (ToM) [30]. In patients with schizophrenia, a study observed that improvement in executive function predicted improvement of social functioning [31]. Although the relationship between cognitive function and social cognition is a topic of great importance in research and clinical practice, it has not been sufficiently addressed. More definitive conclusions from studies would be useful for customized treatment [32].
As both cognitive function and mentalization can influence treatment response in MDD, and mentalization can be influenced by cognitive function, examining the mediating effect of mentalization on the process by which cognitive function influences treatment response would clarify the relationship between mentalization and cognitive function. To our knowledge, most studies [13,14,20,21] have focused on whether cognitive function or mentalization influences treatment response; however, no studies have examined the mediating effect of mentalization between baseline cognitive function and treatment response in MDD. Therefore, we investigate whether cognitive function and mentalization can predict treatment response, and whether mentalization mediates the effect of cognitive function on treatment response. We hypothesized that: 1) A positive correlation was observed between cognitive function and mentalization; 2) Cognitive function and mentalization predicted treatment response after 8 weeks of treatment; and 3) Mentalization mediates the effect of cognitive function on treatment responses.
METHODS
Participants
Twenty-eight adult patients who visited the Department of Psychiatry, Korea University Anam Hospital, Seoul, between September 2021 and September 2022 voluntarily participated in this study. They met psychiatrists’ diagnosis of MDD using the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition [33] and were in the MDE state at baseline according to a comprehensive psychological evaluation. Patients with unipolar MDD without psychotic symptoms were included in the study. Those who met the following criteria were excluded: cerebrovascular disease on brain imaging, head trauma, history of electroconvulsive therapy in the last 6 months, comorbid mental disorders. Written informed consent was obtained from thirty patients; however, two patients who refused the second measurement were excluded. The study was approved by the Institutional Review Board of Korea University Anam Hospital (approval number: 2021AN0428).
Procedures
As this naturalistic study was not designed to test the effectiveness of certain drugs, we did not control for the type of medication taken in 8 weeks; however, we did not combine psychotherapy to limit it to a single therapeutic approach. Regarding medication, nine patients took antidepressants+ anxiolytics, and five patients took antidepressants+atypical antipsychotics, thirteen patients took antidepressants+anxiol ytics+atypical antipsychotics, and one patient was drug-naive at baseline and started on antidepressant medication. All patients were assessed at baseline and at the end of the 8-week treatment period. Cognitive function, mentalization, symptom severity, and functional impairment were measured at baseline, and symptom severity, functional impairment were measured repeatedly after 8 weeks of treatment. Treatment response was defined as the rate of improvement in symptom severity and functional impairment [(baseline score–8-week score)/baseline score]. Using the rate of improvement provided a more accurate assessment of the effect of symptom severity at baseline rather than simply using a score difference. All assessments were conducted by a clinical psychologist and clinical psychology trainees.
Measures
Comprehensive cognitive function was measured using the following instruments: attention/concentration (digit span forward, Arithmetic of Korean-Wechsler Adult Intelligence Scale-IV [K-WAIS-IV]) [34], processing speed (Symbol Search and Coding of K-WAIS-IV), memory (Korean-Auditory Verbal Learning Test) [35], executive function (Digit Span backward of K-WAIS-IV, Stroop, Verbal Fluency, and Design Fluency of Kims Frontal-Executive Function Neuropsychological Test-ll) [36]. All cognitive measurements yielded scaled scores compared with normative data.
Mentalization was measured using Self-Rated Mentalization Questionnaire (SRMQ) [37]. This is a 5-point Likert scale developed in Korea based on Fonagy’s definition of mentalization as a reflective functioning. This 25-item scale has four factors (reflection of the self and others, deficit of emotion awareness, absolute certainty about others’ mind, and concrete thinking). SRMQ shows strong internal consistency (Cronbach’s α=0.85) and good convergent validity with the other mentalization scale (r=-0.53, p<0.01) [37]. The overall scores were used in the data analysis.
We used a self-rated or clinician-rated instrument to assess functional impairments. The scores on the Korean version of the Sheehan Disability Scale (K-SDS) [38], a reliable and valid self-report questionnaire that measures functional impairment in occupation, interpersonal relationships, and family life, were analyzed. The total score was the sum of three item scores (0 to 10 per item). K-SDS shows internal consistency (Cronbach’s α=0.89) and the coefficient of 1 week test-retest reliability (0.84) [38]. Regarding to validity, the total score of KSDS was significantly different between psychiatric group (mean 17.18) and normal group (mean 9.13) [38]. Additionally, the total score of K-SDS was significantly correlated with that on the Global Assessment of Functioning (GAF) [39] (r=-0.29, p<0.01) [38]. GAF that clinicians use to rate the patient’s level of interpersonal, occupational, and academic functioning on a scale from 1 to 100 was used to assess general functioning. According to the study applied to Korean psychiatric patients, GAF has high interrater reliability (r=0.91, p<0.01) and adequate concurrent validity with SDS total score (r=-0.53, p<0.01) [39].
Regarding symptom severity, the Korean version of the Hamilton Depression Rating Scale (K-HDRS) [40], proven to be a valid instrument for the Korean population, was used. This scale consists of 17 items which a clinician rates according to severity of depressive symptoms. K-HDRS shows good internal consistency (Cronbach’s α=0.76), interrater reliability (r=0.94, p<0.001) and convergent validity with Clinical Global Impression Scale [41], Beck Depression Inventory [42], Montgomery-Asberg Depression Rating Scale43 (r=0.84, r=0.54, r=0.58, p<0.001, respectively) [40]. The Korean version of the Center for Epidemiological Studies-Depression Scale (CES-D) [44] was also used. This is a 20-item, 4-point Likert self-report scale. CESD among Korean population shows very good internal consistency (Cronbach’s α=0.91). As a result of factor analysis to test validity, this scale proved to have four-factor structure which explained 62.3% of the total variance [44].
Data analyses
In the first step, Pearson’s correlation coefficients among the variables were used to explore the association between cognitive function and mentalization. In the second step, we conducted multiple regression analyses to explore whether cognitive function and mentalization predicted treatment response after 8 weeks of medication. The treatment response as a dependent variable included four variables: the rate of improvement in symptoms (CES-D, K-HDRS) and functional impairment (K-SDS, GAF). In the first regression analysis, to test the predictive power of cognitive function, four dependent variables were entered separately and eleven indices yielded from cognitive measures were entered as independent variables, one index at a time to avoid multicollinearity between indices. To identify covariates that may influence treatment response, Pearson’s correlation coefficients between each dependent variable with demographic and clinical data were calculated. Demographic and clinical variables with statistical significance were then entered into the regression model with predictors. To test whether mentalization predicted the treatment response, the SRMQ score was entered as an independent variable.
Finally, we conducted an exploratory mediation analysis using the PROCESS macro [45] to check whether mentalization mediates the effect of cognitive function on treatment response. Bootstrapping with 5,000 samples and a 95% confidence interval (CI) was set, and statistical significance was determined if the lower and upper limits of the CI did not contain 0. Bootstrapping is considered an alternative to traditional mediation testing methods because of the high statistical power of mediating effects [45]. All data were analyzed using IBM Statistical Package for the Social Sciences (SPSS) version 23 (IBM Corp.). Statistical significance was set at 0.05.
RESULTS
The demographic and clinical data of the participants are presented in Table 1. The sample comprised twenty-two outpatients and six inpatients. Depressive symptoms were severe on the K-HDRS (24.96) and CES-D (39.96). Regarding functional impairment, the SDS score (21.29) was severely impaired compared with the mean score in the normal control group [38]. The GAF score was 55.39, which was in the range of moderate functional impairment.
Demographic and clinical data (N=28)
Correlation between cognitive function and mentalization
The correlations, means, and standard deviations (SD) of cognitive function and mentalization at baseline are shown in Table 2. The performance of all cognitive measures was calculated using age-adjusted scaled scores (mean 10, SD 3), within the normal range (scores 7–13), indicating no cognitive impairment. As a result of correlational analyses between mentalization and cognitive function, SRMQ scores showed significant correlations with design fluency (r=0.58, p<0.001), immediate recall (r=0.49, p=0.01), delayed recall (r=0.55, p<0.001), recognition (r=0.48, p=0.01), Digit Span forward (r=0.39, p=0.04), and backward (r=0.50, p=0.01). There was no correlation between the SRMQ scores and other cognitive measures.
Correlation coefficients, means, and SD of variables
Cognitive function and mentalization as predictors of treatment response
Because a correlation was observed between mentalization and some domains of cognition in this study, regression analyses were conducted before and after controlling for mentalization. Current age and age at onset were entered as covariates, as these variables were significantly correlated with treatment response in this study. Correlational analyses revealed no association between any of the cognitive measures and treatment responses. Therefore, regression analyses yielded no significant results. In the regression analyses controlling for mentalization, however, delayed recall significantly predicted improvement in GAF (B=-0.03, p=0.02). The same analytic procedure was used to investigate the effect of mentalization as a predictor of treatment response. Similarly, no significant results were observed in correlation analyses. Regression analyses controlling for cognitive function showed that mentalization significantly predicted the rate of improvement in the GAF after controlling for delayed recall and Digit Span backward (B=0.01, p=0.02; B=0.01, p=0.04, respectively).
Mediating effect of mentalization
Some cognitive functions predicted the treatment response after controlling for mentalization, and conversely, mentalization predicted the treatment response after controlling for cognitive functions, raising the need to verify the mediating effect of mentalization. Mediational analyses showed that the indirect effect of mentalization on the relationship between cognitive measures (delayed recall and Digit Span backward) and treatment response was significant, as the lower and upper limits of CI did not contain 0. The results are presented in Tables 3, 4, and Figure 1. As shown, the direct effect of delayed recall on functional improvement (GAF) was significant (B=-0.033, p=0.02) as well as the mediating effect of mentalization (B=0.019). Considering the direction of the mediating effect, as the delayed recall scores increased, so did the SRMQ score (B=1.624, p<0.001), and an increased SRMQ score had a positive effect on functional improvement (B=0.012, p=0.02). When age at onset (r=0.44, p=0.02) was entered as a covariate, the direct effect remained significant, whereas the indirect effect did not. The direct effect of Digit Span backward on functional improvement (GAF) was not significant (B=-0.026, p=0.062), whereas the mediating effect of mentalization was significant (B=0.014). Regarding the path, the increasing Digit Span backward scores increased the SRMQ score (B=1.435, p=0.01), and the increased SRMQ score had a positive effect on functional improvement (B=0.01, p=0.04). However, when the age of onset was controlled, the mediating effect vanished. Regarding to mediation model’s goodness of fit, total effect of delayed recall and Digit Span backward on functional improvement was not statistically significant (F=1.30, p=0.26; F=0.93, p=0.34, respectively). These results are due to the different direction of the direct and indirect effects and do not eliminate the need to test mediation effect. This is known as suppression effect in mediation model, which discussed in the current study [46-48].
Mediating effect of mentalization 1
Mediating effect of mentalization 2
Graphics of mediating effects. GAF, Global Assessment of Functioning. *p<0.05; **p<0.01.
DISCUSSION
This prospective study investigated the predictive power of cognitive function and mentalization at baseline for 8-week treatment responses and tested the mediating role of mentalization on the effects of cognitive function.
Significant positive correlations were observed between mentalization and various cognitive functions, supporting the hypotheses of this study. According to the results, the better the memory, executive function, and attention, the higher the mentalizing ability. This suggests that not only simple attention but also higher-order cognitive functions such as memory, working memory, cognitive flexibility, and self-regulation may be associated with mentalization. Stroop, symbol search, and coding among the cognitive measures were not significantly related to mentalization, suggesting that cognitive flexibility rather than cognitive inhibition may be directly related to mentalization, and processing speed may not be related to mentalization. This result is in line with those of studies that reported positive correlations between mentalization and executive function [20,24]. Given the lack of associations [28,29] and the lack of literature directly addressing the relationship between mentalization and cognitive function in patients with MDD, the results of this study may provide important information for comparison in future studies.
Regression analysis, controlling for mentalization and cognitive function, showed that certain cognitive functions and mentalization predicted treatment response. Delayed recall was the only cognitive index that predicted the functional improvement. The possibility that patients with lower memory test retrieval will show greater functional improvement after short-term medication needs to be verified in further studies with larger samples. This study’s results suggest that most cognitive measures, except delayed recall, did not predict treatment response, which may be a consequence of the younger sample (mean age=23 years). According to a systematic review [14], executive function was the best predictor of treatment response in MDD [49-52], whereas learning and memory were not significantly associated with treatment response [53-55]. This result is consistent in older patients [56-58], but inconsistent in younger patients, making it difficult to draw conclusions [14]. For example, some findings suggest that patients with poor memory [59], attention, and working memory [50] may have a poor treatment response, but there are also reports that cognitive function does not predict treatment response [60,61] or that patients with lower executive function show better responses [62,63]. The finding that mentalization predicts functional improvement after controlling for cognitive function suggests that mentalization may directly influence functioning, such as interpersonal relationships, rather than depressive symptom improvement. This can explain why the capacity to mentalize plays a substantial role in interpersonal relationships. Longitudinal studies with longer follow-up periods (6 months and 1 year) are needed to explore whether mentalization reduces interpersonal discomfort, which in turn reduces depressive symptoms.
The hypothesis that mentalization mediates the effect of cognitive function on treatment response was supported; however, the mediating effect lost significance after controlling for covariates (age at onset). Although the mediating effect disappeared, the results of the study showed an inconsistent mediation model with a suppression effect [46-48]. This means that the relationship between cognitive function and treatment response was suppressed when mentalization was absent in the model, suggesting that cognitive function has no effect on treatment response. However, the suppressed pathway between cognitive function and treatment response was revealed when mentalization was included in the model.
With regard to the direct effect of delayed recall on treatment response, lower delayed recall scores at baseline were associated with functional improvements after treatment. This finding suggests that patients with poorer memory may benefit more from short-term medication. Since delayed recall requires both memory and executive function [36], the negative effect of delayed recall on treatment response could be considered in light of the above studies [62,63] that report showing lower executive function was associated with better treatment response. A study comparing the cognitive function of responders and nonresponders to selective serotonin reuptake inhibitors (SSRIs) reported that responders were characterized by higher performance on simple tasks and lower performance on complex tasks requiring more cognitive effort [64]. This suggests that patients with deficits in executive function may benefit more from treatment with SSRIs. This result raises the possibility that medications (SSRIs and serotonin-norepinephrine reuptake inhibitors) affect the serotonin and norepinephrine systems in the prefrontal regions involved in executive function, leading to a better treatment response while improving executive function [65].
Our study has several limitations. Despite the prospective investigation, the results should be interpreted with caution because of the small sample size and multiple analyses with various cognitive measures as independent variables, which may have led to a type I error. In a larger sample, repetitive cognitive measurements at follow-up are recommended to identify whether improvements in executive function may have influenced treatment response. This was a naturalistic study in which clinicians prescribed and titrated medications based on the patient’s condition. This increases the ecological validity of the results as it reflects a real-world clinical setting. However, given that the predictability of cognitive function in treatment response may differ depending on the antidepressant mechanism (SSRI vs. norepinephrine-dopamine reuptake inhibitor) [62], comparisons between groups by medication type are suggested to obtain more robust results.
The clinical implication of our study is that it shows that a certain cognitive function may influence pharmacological treatment response in different ways. Specifically, good cognitive resources may not increase treatment response directly, but rather influence through social cognition, such as mentalization, that improves global functioning. Applying this to treatment of MDD, for patients without cognitive deficits (memory or working memory), a combination with psychotherapy such as mentalization-based treatment from the beginning is expected to increase pharmacological treatment response.
Notes
Availability of Data and Material
The datasets generated or analyzed during the study are available from the corresponding author on reasonable request.
Conflicts of Interest
The authors have no potential conflicts of interest to disclose.
Author Contributions
Conceptualization: Seon Hee Hwang. Methodology: Seon Hee Hwang. Data curation: Seon Hee Hwang, Byung Joo Ham. Formal analysis: Seon Hee Hwang. Investigation: Seon Hee Hwang. Supervision: Myung Sun Kim. Writing—original draft: Seon Hee Hwang. Writing—review & editing: Myung Sun Kim, Byung Joo Ham.
Funding Statement
This research was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant number: HI23C1234).
Acknowledgments
None