Association between treatment resistance and cognitive function in schizophrenia

Abstract

Introduction:

Treatment-resistant schizophrenia (TRS) affects around 30% of individuals with schizophrenia. About half of the patients with TRS who are treated with clozapine do not show a meaningful clinical response, that is, clozapine resistance. To date, the relationship between cognitive function and treatment response categories is not entirely clear. This study evaluated the cognitive performance across subgroups stratified by treatment response, and we hypothesised that cognitive impairment increases with increased treatment resistance.

Methods:

This study was conducted at the Institute of Mental Health, Singapore, and included healthy controls and people with schizophrenia categorised into these groups: antipsychotic-responsive schizophrenia (ARS), clozapine-responsive TRS (TRS-CR) and clozapine-resistant TRS (ultra-treatment-resistant schizophrenia [UTRS]). Cognitive function was assessed using the Brief Assessment of Cognition-Short Form. Symptoms were measured with the Positive and Negative Syndrome Scale (PANSS). The planned statistical analyses included adjustments for covariates such as age, sex, PANSS scores and antipsychotic dose, which might affect cognitive function.

Results:

There were significant differences in overall cognitive performance between the groups: ARS had the least impairment, followed by TRS-CR and UTRS. Antipsychotic dose, and PANSS negative and disorganisation/cognitive factors were significant predictors of overall cognitive function in all patient groups.

Conclusions:

Our study found differences in cognitive function that aligned with levels of treatment resistance: the greater the degree of treatment resistance, the poorer the cognitive function. Interventions to improve negative and disorganisation symptoms might be effective to enhance the cognitive function and treatment outcomes in schizophrenia.

INTRODUCTION

Treatment-resistant schizophrenia (TRS) affects approximately 30% of patients diagnosed with schizophrenia.[1] It has been reported that differences exist between TRS and non-TRS patients in terms of neurochemical and structural abnormalities, and this suggests that TRS is a clinical and/or pharmacological subtype.[2] Clozapine is the only antipsychotic with established efficacy in symptoms and relapse reduction in TRS.[3] Around 40% of patients with TRS have unsatisfactory responses despite an adequate trial of clozapine, and this condition is known as clozapine-resistant schizophrenia,[4] or ultra-treatment-resistant schizophrenia (UTRS).

Cognitive impairment is a core feature of schizophrenia,[5] particularly in the domains of processing speed, verbal memory and working memory.[6] Cognitive deficits are significantly linked to poor functional outcomes[7] and are an independent domain from psychotic symptoms.[8] A 10-year follow-up study has shown that early stabilisation may predict better long-term neurocognitive functioning.[8]

SUMMARY BOX

What is known?

Treatment-resistant schizophrenia (TRS) affects a substantial number of schizophrenia patients, with a proportion of patients being ultra-treatment resistant. Features of treatment resistance manifest in differences in cognitive impairments between subgroups of patients.

What is new?

This study found worsening cognitive impairments across increasing levels of treatment resistance, that is, antipsychotic-responsive schizophrenia, clozapine-responsive TRS and ultra-treatment-resistant schizophrenia. It also identified antipsychotic dose, and PANSS negative and cognitive/disorganisation factor scores as significant predictors of cognitive impairment across these subgroups.

What is the impact?

The findings highlight that targeting negative and disorganisation symptoms might enhance the cognitive function and treatment outcomes in schizophrenia.

Studies have shown that TRS patients exhibit more robust cognitive impairments across several domains, such as selective attention, cognitive flexibility, processing speed, executive functions, verbal fluency[9] and verbal memory,[10] as compared to non-TRS patients. However, there is limited research on comparison of performances between the pharmacological subtypes — antipsychotic-responsive schizophrenia (ARS), clozapine-responsive TRS (TRS-CR) and UTRS.[2] A previous study showed similar cognitive performance between the three group of patients,[11] while another study found greater severity of treatment resistance associated with greater impairments in sustained attention.[12] Spangaro et al.[13] reported that patients with UTRS are associated with overall longitudinal cognitive decline, whereas clinical stabilisation and adequate antipsychotic response in ARS and TRS-CR patients are linked to moderate global cognitive improvement over time.

Taken together, there has been mixed evidence on the relationship between response to treatment and cognitive performance in patients with schizophrenia, in particular, the differences between TRS-CR and UTRS. Moreover, existing research was conducted mainly in Caucasian populations and had small sample sizes. This study aimed to compare the differences in cognitive performance between patients with ARS, TRS-CR and UTRS. We hypothesised that a greater degree of treatment resistance is linked to poorer cognitive function.

METHODS

Participants

This study was conducted at the Institute of Mental Health, Singapore, the only tertiary mental health institution in the country providing a full range of psychiatric services, including clozapine outpatient clinics for patients with TRS. A total of 150 participants with a diagnosis of schizophrenia and 50 healthy volunteers, aged 21–69 years, were recruited between July 2019 and September 2022. Participants with schizophrenia were categorised into these groups: ARS (n = 50), TRS-CR (n = 70) and UTRS (n = 30). Two patients in the ARS group were excluded from analysis owing to incomplete data.

The ARS participants were in positive symptom remission on first- or second-line antipsychotic agents. All participants identified as either TRS-CR or UTRS had at least two previous unsuccessful antipsychotic trials of minimally 6-week duration and >8 weeks of clozapine treatment.[14] Participants with TRS-CR had attained positive symptom remission with clozapine. Participants with UTRS had not attained positive symptom remission despite an adequate dose (serum clozapine level >350 ng/mL).

Ethics approval for this study was provided by the National Healthcare Group’s Domain Specific Review Board (reference: 2018/00993), and informed consent was obtained from all study participants before study assessments.

Clinical assessment

Diagnosis of schizophrenia was ascertained on the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, 4^th Edition, Text Revision (DSM-IV-TR) by trained and experienced clinicians. Positive symptom remission was defined by a Clinical Global Impression-Schizophrenia positive symptom item score ≤3, as rated by trained clinicians. All groups of patients were required to have no changes to their current prescription for the past 2 weeks. Individuals selected as controls were healthy community-dwelling individuals with no history of mental illness or neurological disorder.

Specific psychopathology was rated on the Positive and Negative Syndrome Scale (PANSS), a 30-item instrument widely used in the assessment of specific and general psychiatric symptoms. The PANSS items are rated on a 7-point scale (1 = absence of symptoms, 7 = extremely symptomatic). The five-factor PANSS structure (positive, negative, cognitive/disorganisation, depression/anxiety and hostility) [see Supplemental Digital Appendix], which was validated locally, was applied to quantify the symptom dimensions.[15]

Neurocognition was evaluated on the Brief Assessment of Cognition-Short Form (BAC-SF), a brief three-task cognitive screener developed from the Brief Assessment of Cognition in Schizophrenia (BACS) that has been found to robustly summarise overall cognitive performance in schizophrenia.[16] The BAC-SF consists of digit sequencing, symbol coding and verbal memory, which measure working memory, processing speed and short-term verbal memory, respectively. The composite score obtained from the three tasks has been shown to agree strongly with BACS (r = 0.91). Antipsychotic doses prescribed for patients were converted to total daily chlorpromazine (CPZ) equivalents.[17,18]

Statistical analysis

Analysis of variance was used to compare the clinical and demographic characteristics of the ARS, TRS-CR and UTRS groups. If there was a statistically significant association, the variable was entered as a covariate when comparing group differences. Analysis of variance was also used to compare cognitive function between the groups. To better explore the impact of specific predictors on cognitive outcomes, linear regression was performed with potential confounding factors such as age, sex, PANSS score and daily CPZ dose, entered as covariates in linear regression analysis to adjust for their effects. Analysis of each of the five PANSS factors was done to examine the influence of specific symptom domains on cognitive function in treatment groups. A collinearity analysis was done to ensure the independence of predictor variables. During this analysis, the PANSS positive factor was identified as having a tolerance value <0.6, indicating potential issues with multicollinearity, and was hence excluded from subsequent analyses. All statistical analyses were carried out using IBM SPSS Statistics version 23 (IBM Corp, Armonk, NY, USA). In all statistical analyses, a P value <0.05 was considered statistically significant, indicating <5% probability that the observed differences occurred by chance.

RESULTS

Sample characteristics are presented in Table 1. While healthy controls had a significantly lower age and more years of education compared to patients with schizophrenia, there were no significant differences among treatment groups. On comparing the treatment groups, a statistically significant difference was found in the PANSS total score (P < 0.001), with ARS having the lowest score followed by TRS-CR and UTRS. Mean antipsychotic dose in daily CPZ equivalents increased with increasing treatment resistance. Tukey post hoc analysis showed that total daily antipsychotic dose was significantly higher in UTRS patients as compared to ARS and TRS-CR patients. There were no significant differences between ARS and TRS-CR groups.

Table 1.

Demographics and clinical characteristics of the study sample.

Variable	n (%)/mean±SD				P
	Healthy controls (n=50)	ARS (n=48)	TRS-CR (n=70)	UTRS (n=30)
Gender
Male	27 (54)	25 (52)	41 (59)	20 (67)	0.605
Female	23 (46)	23 (48)	29 (41)	10 (33)
Ethnicity
Chinese	44 (88)	34 (71)	59 (84)	27 (90)	0.106
Malay	4 (8)	6 (12)	4 (5)	1 (3)
Indian	1 (2)	7 (14)	6 (8)	2 (7)
Others	1 (2)	1 (2)	1 (1)	0 (0)
Mean age (yr)	34.68	39.50	37.94	41.23	0.047
Average education (yr)	15.16	13.20	12.61	13.00	<0.001
Antipsychotic dose in CPZ equivalents (mg)	–	297.67±280.30	309.14±259.35	478.29±385.32	0.018*
PANSS total score	34.26±3.43	43.46±7.87	49.80±10.34	67.27±11.87	<0.001*
BAC-SF composite score	–0.07±0.66	–1.06±0.96	–1.38±0.97	–1.73±1.11	0.017*
BAC-SF verbal memory	–0.26±1.19	–1.24±1.40	–2.02±1.40	–1.94±1.44	0.011*
BAC-SF digit sequencing	0.24±0.98	–0.92±1.19	–0.89±1.33	–1.38±1.34	0.191*
BAC-SF symbol coding	0.23±0.83	–1.03±0.97	–1.22±0.90	–1.86±1.23	0.002*

*P-value for comparisons between treatment groups only. ARS: antipsychotic-responsive schizophrenia, BAC-SF: Brief Assessment of Cognition-Short Form, CPZ: chlorpromazine, PANSS: Positive and Negative Syndrome Scale, SD: standard deviation, TRS-CR: clozapine-responsive treatment-resistant schizophrenia, UTRS: ultra-treatment-resistant schizophrenia

Significant group differences were noted between healthy controls and patients with schizophrenia in all categories assessed on BAC-SF, with varying degrees of differences observed between groups [Figure 1]. Notably, the BAC-SF composite scores differed significantly across the treatment groups (P = 0.017), where ARS patients had the best performance, followed by TRS-CR patients and UTRS patients. Post hoc comparisons using Tukey’s test indicated significant disparities between ARS and UTRS patients (P = 0.013).

Figure 1.

Graph shows the Brief Assessment of Cognition-Short Form (BAC-SF) scores across groups. ARS: antipsychotic-responsive schizophrenia, TRS-CR: clozapine-responsive treatment-resistant schizophrenia, UTRS: ultra-treatment-resistant schizophrenia

For BAC-SF verbal memory, there are significant differences between the different treatment groups (P = 0.011), with ARS patients having the highest score, followed by TRS-CR and UTRS patients. Post hoc analysis showed significant differences between ARS and TRS-CR patients (P = 0.010). For BAC-SF digit sequencing, there were no significant differences between treatment groups. Similarly, the BAC-SF symbol coding test showed significant differences between treatment groups (P = 0.002). Significant differences were particularly evident between ARS and UTRS patients (P = 0.001), as well as between TRS-CR and UTRS patients (P = 0.010), while no significant differences were observed between ARS and TRS-CR patients.

Table 2 shows the relationship between the different predictors and BAC-SF cognitive composite score. Higher antipsychotic doses and higher PANSS negative and cognitive/disorganisation factor scores were associated with overall cognitive performance. The PANSS cognitive/disorganisation factor was consistently associated with cognitive function across all three BAC-SF tasks [Table 3]. In addition, antipsychotic dose and PANSS negative and depression/anxiety factors emerged as significant predictors of BAC-SF verbal memory. Antipsychotic dose was a significant predictor of BAC-SF symbol coding.

Table 2.

Linear regression for Brief Assessment of Cognition-Short Form composite score.

Predictor	Coefficient	Standard error	t	P
Gender	–0.188	0.149	–1.260	0.210
Age	–0.003	0.007	–0.420	0.675
Subject group	–0.056	0.119	–0.470	0.639
Antipsychotic dose in CPZ equivalents	–0.001	0.000	–2.254	0.026*
PANSS negative factor	–0.038	0.018	–2.087	0.039*
PANSS cognitive/disorganisation factor	–0.128	0.028	–4.592	<0.001*
PANSS depression/anxiety factor	0.033	0.034	0.972	0.333
PANSS hostility factor	–0.010	0.046	–0.219	0.827

R² = 0.316, P for model <0.001. *Statistically significant. CPZ: chlorpromazine, PANSS: Positive and Negative Syndrome Scale

Table 3.

Linear regression for Brief Assessment of Cognition in Schizophrenia (BACS) verbal memory, digit sequencing and symbol coding.

Predictor	BACS verbal memory				BACS digit sequencing				BACS symbol coding
	Coefficient	SE	t	P	Coefficient	SE	t	P	Coefficient	SE	t	P
Gender	–0.499	0.212	–2.350	0.020	0.136	0.207	0.658	0.512	–0.201	0.158	–1.271	0.206
Age	–7.312E-5	0.010	–0.007	0.994	–0.011	0.010	–1.197	0.233	0.003	0.007	0.389	0.698
Subject group	–0.128	0.170	–0.756	0.451	0.143	0.166	0.861	0.391	–0.183	0.127	–1.441	0.152
Antipsychotic dose in CPZ equivalents	–0.001	0.000	–3.114	0.002*	-5.376E-5	0.000	–0.144	0.886	–0.001	0.000	–2.007	0.047*
PANSS negative factor	–0.058	0.026	–2.258	0.026*	-0.021	0.025	–0.834	0.406	–0.034	0.019	–1.780	0.077
PANSS cognitive/disorganisation factor	–0.137	0.040	–3.445	0.001*	-0.167	0.039	–4.317	<0.001*	–0.080	0.030	–2.714	0.007*
PANSS depression/anxiety factor	0.102	0.048	2.101	0.037*	–0.012	0.047	–0.252	0.801	0.009	0.036	0.259	0.796
PANSS hostility factor	0.000	0.066	–0.007	0.994	–0.010	0.064	–0.159	0.874	–0.020	0.049	–0.401	0.689

R²: BACS verbal memory = 0.295, BACS digit sequencing = 0.193, BACS symbol coding = 0.241. *Statistically significant. CPZ: chlorpromazine, PANSS: Positive and Negative Syndrome Scale, SE: standard error

DISCUSSION

This study aimed to delineate differences in cognitive performance in people with schizophrenia grouped by antipsychotic response status — ARS, TRS-CR and UTRS. Our study found significant differences in cognitive performance among ARS, TRS-CR and UTRS, indicating a relationship between antipsychotic response status and cognition.

Persistent deficits in verbal memory represent a core aspect of cognitive impairment in schizophrenia. Notably, differences in impairment of verbal domains have been consistently reported in previous research.[10,19] They have been observed between TRS patients and those responsive to antipsychotic treatment from the initial episode of illness.[20] Our findings contribute to the ongoing discussion by further substantiating the notion that deficits in verbal domains may serve as a cognitive hallmark of TRS. However, the relatively small UTRS sample might have affected the study’s ability to detect statistically significant differences between UTRS and other antipsychotic groups.

In our analysis, we included pertinent covariates for illness outcomes, such as age, sex, antipsychotic dose and severity of psychopathology, to account for potential confounding factors. After controlling for covariates, there were no statistically significant differences among the treatment groups for all three cognitive measures. This finding supports a previous report comparing cognitive function between antipsychotic response groups, and controlled for disease duration and PANSS score.[11] This also highlights the influential role of relevant clinical covariates in the investigation between treatment response and cognitive function in schizophrenia. While our adjusted analyses suggest that differences in cognitive function between groups can be largely attributed to the covariates considered, the identification of verbal memory deficits in TRS across literature and our small UTRS sample size suggest that further research with more robust sampling is necessary to fully elucidate the cognitive profiles characteristic of each treatment group.

The significant inverse relationship observed between PANSS cognitive/disorganisation factor scores and all cognitive tasks in our study is consistent with the findings documented in extant literature, which reported that higher levels of disorganisation symptoms are associated with poorer performance on neuropsychological tests.[21,22] This is congruent with neuroimaging studies that highlighted the involvement of dorsal lateral prefrontal cortex (DLPFC) dysfunction in both cognitive impairment and behavioural disorganisation.[23] Consequently, targeting DLPFC dysfunction may represent a promising therapeutic strategy for alleviating disorganisation symptoms and improving cognitive outcomes.

Notably, our analysis revealed a significant inverse relationship between PANSS negative factor scores and cognitive performance. This is concordant with previous research findings.[10] Longitudinal studies have shown that patients with more pronounced negative symptoms tended to demonstrate poorer global functioning, with cognitive composite score exerting an independent effect, underscoring the importance of effectively treating negative symptoms in schizophrenia.[24] Thus, our study highlights the intricate interplay between negative symptoms and cognitive deficits in schizophrenia; both domains remain key unmet clinical needs, and effective treatment strategies are sorely lacking.

Interestingly, our study found a significant positive association between the PANSS depression/anxiety factor and BAC-SF verbal memory. Depressive symptoms have been linked to poorer cognitive performance in younger adults with schizophrenia,[25] but better cognitive performance in older adults.[26] This highlights the substantial heterogeneity in schizophrenia, where the relationship between specific symptom dimensions and cognitive function may vary between subgroups or stages of disease. Earlier studies relating anxiety levels to cognitive function also showed an inverted U-shaped relationship.[27] This phenomenon aligns with the Yerkes–Dodson law, which suggests that an increase in anxiety may temporarily enhance cognitive performance.

Our study observed a significant negative relationship between antipsychotic dose and BAC-SF composite score, notably in verbal memory and symbol coding scores. While we hypothesised that anticholinergic burden may explain the cognitive impairments as per previous research,[28] our study did not directly measure anticholinergic burden. We used CPZ equivalent doses as a marker for overall antipsychotic exposure, which may also reflect anticholinergic burden due to the anticholinergic properties of some antipsychotics. This highlights the importance of individualising antipsychotic regimens to mitigate polypharmacy and minimise anticholinergic burden, thereby potentially ameliorating cognitive impairments.

There are some limitations in our study. The cross-sectional nature of the study limits our ability to infer causality or assess the dynamic trajectories of cognitive impairment and symptom severity over time. Future longitudinal studies could address this limitation by tracking changes in cognitive function and symptomatology longitudinally, providing a more comprehensive understanding on the inter-relationships. The reliance on self-report measures for certain constructs, such as symptom severity and medication adherence, introduces the possibility of recall bias and social desirability bias. Incorporating objective measures or multi-method assessments might provide a more accurate representation in future research. It is important to note that CPZ equivalents do not directly quantify anticholinergic activity, as different antipsychotics have varying degrees of anticholinergic properties. Future studies should incorporate specific anticholinergic burden scales to provide a more accurate assessment of the impact of anticholinergic properties on cognitive function.

Despite these limitations, our findings underscore the importance of further investigating the complex interplay among cognitive impairment, symptomatology and treatment outcomes in TRS, with potential implications for personalised treatment approaches and intervention strategies aimed at improving functional outcomes in TRS patients. Future research should continue to explore novel avenues for intervention, including cognitive remediation strategies, psychosocial interventions and pharmacological approaches, to address the challenges associated with TRS.

In conclusion, our study found that antipsychotic resistance is associated with poorer cognitive function. Antipsychotic doses, disorganisation and negative symptoms play a significant role in this relationship. Developing effective interventions to target specific symptom clusters, such as disorganisation and negative symptoms, or overall antipsychotic load holds promise for optimising cognitive and treatment outcomes in schizophrenia.

Financial support and sponsorship

Nil.

Conflicts of interest

Yee JL had received honoraria and served as a consultant or advisory board member in Otsuka, Janssen, Lundbeck, Sumitomo Pharmaceuticals, Boehringer Ingelheim and ThoughtFull World Pte. Ltd.

Supplemental digital content

Appendix at http://links.lww.com/SGMJ/A142

APPENDIX

Table S1.

PANSS factor and items

PANSS factor	PANSS items
Positive symptoms factor	P1 Delusions P3 Hallucinatory Behaviour P5 Grandiosity P6 Suspiciousness/persecution G9 Unusual thought content G12 Lack of judgement and insight
Negative symptoms factor	N1 Blunted affect N2 Emotional withdrawal N3 Poor rapport N4 Passive/apathetic social withdrawal N6 Lack of spontaneity and flow of conversation G7 Motor retardation G16 Active social avoidance
Cognitive/disorganisation factor	P2 Conceptual disorganisation N5 Difficulty in abstract thinking N7 Stereotyped thinking G5 Mannerism & posturing G10 Disorientation G11 Poor attention G13 Disturbance of volition
Anxiety/depression factor	G2 Anxiety G3 Guilt feelings G6 Depression
Hostility factor	P4 Excitement P7 Hostility G4 Tension G8 Uncooperativeness G14 Poor impulse control