Clinical Predictors of Response to Clozapine in Patients with Treatment Resistant Schizophrenia

Abstract

Objectives

Despite clozapine’s superior clinical efficacy in Treatment Resistant Schizophrenia (TRS), its adverse effects, need for periodic leukocyte monitoring, cost and variable clinical outcomes make the therapeutic decision making process difficult and mandate a clinical need to predict its treatment response. Hence, we investigated various clinical variables associated with treatment responses and adverse events of clozapine in TRS.

Experimental Design

We assessed socio-demographic and clinical profiles, premorbid adjustment, traumatic life events, cognition, disability, psychopathology and serum clozapine levels of 101 patients with TRS on stable dose of clozapine using the following instruments: Brief Psychiatric Rating Scale, Abnormal Involuntary Movements Scale, Addenbrooke’s Cognitive Examination—Revised, WHO Disability Assessment Scale-II, Childhood and Recent Traumatic Events Scale, and Premorbid Assessment Scale. We defined clozapine response a priori, adopted a case-control design framework and employed appropriate multivariate analyses.

Principal Observations

Past history of catatonia (p = 0.005), smoking more than one pack/day (p = 0.008), hyper-somnolence (p = 0.03) and cognitive dysfunction (p = 0.007) were associated with non-response to clozapine. Outcome definitions of non-response to clozapine influenced its association with clinical predictors.

Conclusions

Clinical variables are useful to predict response to clozapine. Smoking can be a potentially modifiable risk factor. Future longitudinal studies, investigating clinical and pharmacogenetic variables together, are desired.

Introduction

Clozapine is the drug of choice for the management of Treatment Resistant Schizophrenia (TRS).¹ The advantages of clozapine include its superior clinical efficacy, its ability to reduce negative symptoms as well as the risk for suicide² and its low propensity to cause movement disorders.³ Its disadvantages are suboptimal response in 40–70% patients with TRS,⁴ high cost, serious adverse events, such as seizures, agranulocytosis, weight gain and dyslipidemia,⁵ and the need for periodic monitoring of leukocytes. Therapeutic decision-making process for starting clozapine is difficult and vacillating for many patients, their families and psychiatrists. There is a clinical need to identify the factors associated with response to clozapine in TRS.⁶

Research on the factors associated with clinical outcomes of clozapine in schizophrenia has focussed mainly on genetic^7-10 and other biological markers.^11,12 However, many biological predictors are inconclusive, contradictory or pending replication⁹ and their clinical utility is currently doubtful.¹³ Clinical variables, such as baseline psychopathology and premorbid functioning have been reported as reliable predictors of clozapine response.⁶ Clinical psychiatrists often consider more clinical variables, before discussing the option of clozapine to their patients with TRS.

Most studies evaluating clinical predictors of response to clozapine^14-19 were not specific to TRS and have recruited patients without treatment resistance as well as patients with schizoaffective disorders. They have seldom employed standard instruments to assess premorbid adjustment, traumatic life events, cognition and disability. Many studies have not estimated serum clozapine levels of their participants. Despite ethnic differences in the plasma levels and the clinical responses of clozapine are known to exist,²⁰ pertinent data on the clinical predictors of response to clozapine in Asian patients remain sparse.²¹ Hence, we aimed to investigate the clinical predictors of response to clozapine, exclusively in patients with TRS in India, with structured assessment of various clinical variables and estimation of serum clozapine levels.

Materials and Methods

Study Design

We employed a cross-sectional study to investigate our objectives. We used a case control design framework to identify the clinical variables associated with non-response to clozapine.

Setting

We conducted this study in Department of Psychiatry, Christian Medical College (CMC), Vellore, India, a tertiary referral centre for the management of psychiatric disorders. The hospital has short-term inpatient services, daily outpatient and regular follow-up clinics. Patients with schizophrenia are initially treated with either dopamine antagonists or serotonin dopamine antagonists (SDA). Clozapine is never used as the first line antipsychotic and is reserved for patients with TRS. Leukocyte counts and metabolic parameters are periodically monitored. Detailed medical records of treatment are maintained for all patients. Most of our outpatients with schizophrenia live in the community with their families. Their medications are directly provided by their first degree relatives or spouses, who report any degree of non-adherence to the treating psychiatrists during periodically scheduled follow-up visits.

Recruitment of Participants

We invited all consecutive patients who satisfied the following eligibility criteria: (i) DSM IV-TR diagnosis of schizophrenia (ii) established treatment resistance in the past after failure to respond at least two adequate antipsychotic trials, as documented by treating psychiatrists. An adequate antipsychotic trial was defined by 600 mg chlorpromazine equivalents for duration of at least six weeks with good drug compliance. The two adequate antipsychotic trials included at least one adequate trial with a SDA. (iii) On stable dose regimen of clozapine for at least twelve weeks with good drug compliance during that period. Participants with severe neurological illnesses, intellectual disability and sensory impairment, precluding the assessment, were excluded.

Assessment

We employed the following instruments:

1. Brief Psychiatric Rating Scale (BPRS): BPRS covers a broad range of psychopathology including thought disturbance, emotional withdrawal, anxiety, depression, hostility and suspiciousness. It has good psychometric properties²² and is used widely in clinical settings.

2. Abnormal Involuntary Movements Scale (AIMS): AIMS is a clinical examination and rating scale developed to measure antipsychotic drug induced dyskinesia.²³

3. Addenbrooke’s Cognitive Examination (ACE-R): ACE-R is a brief cognitive test battery, which incorporates the sub-domains, orientation, attention, memory, verbal fluency, language and visuo-spatial. ACE-R has very good internal consistency, convergent validity and is sensitive to early cognitive dysfunction.²⁴

4. WHO Disability Assessment Scale II (WHODAS II): WHODAS II assesses the disability status of physically and psychiatrically ill, by evaluating their functioning in the domains of communication, mobility, self-care, interpersonal, life activities, and participation.^25,26

5. Childhood Traumatic Events Scale (CTES): CTES briefly assesses six early traumatic experiences prior to the age of seventeen such as bereavement, parental discord, violence, sexual abuse, physical illness and other major life events.²⁷

6. Recent Traumatic Events Scale (RTES): RTES assesses seven traumatic experiences within the past three years such as bereavement, divorce, violence, sexual abuse, physical illness, occupational change and other major life events.²⁷

7. Premorbid Assessment Scale (PAS): PAS is a widely used rating scale to reliably assess premorbid functioning retrospectively.²⁸

8. Socio-demographic, clinical and treatment profile: We employed a structured questionnaire to collect socio-demographic, clinical and treatment data. We recorded data about developmental delay, urbanization, caffeine consumption, smoking and anthropometric measures. We collected self reported data on clozapine related adverse effects, using a detailed check list.

ACE-R, CTES, RTES and PAS were translated into the local language, Tamil, and then back translated to English by bilingual health professionals, who focused on content, conceptual, semantic as well as technical equivalence. The final Tamil versions were obtained by consensus among the translators.

Data Collection

Every participant was individually assessed for psychopathology (SB) using BPRS. An independent investigator (CC) employed other instruments and assessed various clinical variables by detailed personal interviews with the participants and their primary care givers. We collected peripheral venous blood sample from all participants, 12 hours after their last clozapine dose. Serum clozapine levels were measured by high performance liquid chromatography with ultraviolet detection.²⁹ Investigators collecting data on outcome (SB) and exposure variables (CC) remained blind to each other’s findings till the completion of the study. The protocol of this study was approved by the Institutional Review Board of CMC, Vellore, India. We provided a fact sheet about the details of this study to all participants. We discussed those details and obtained written informed consent from the participants and from their first-degree relatives or spouses. We accessed the medical records of all participants with their consent.

Data Analyses

We initially analysed the study variables using descriptive statistics. Many studies have defined the response to clozapine by greater than 20% reduction in the total score of BPRS.³⁰ However, most clinical psychiatrists do not refer to non-response based on a change on any rating scale, but rather on the presence of persistent positive or negative symptoms.³¹ Moreover, due to our cross-sectional study design, we defined the response to clozapine, with the widely employed cross-sectional threshold of having BPRS total score of 35 or less.^30,31 We dichotomously categorized the participants, who had BPRS total scores equal to or less than 35, as clozapine responders. We converted other continuous clinical variables to categorical measures, by splitting them into two categories, using their median values.³² We calculated the odds ratios (OR) with 95% confidence intervals for various hypothesised clinical predictors. When a clinical variable was absent in either the responder or in the non-responder group, we employed the Fisher’s exact test to assess its statistical significance, as odds ratios cannot be calculated. We employed multiple logistic regression analyses to calculate adjusted odds ratios (AOR) accounting for the effects of age, gender and serum clozapine levels. We used Nagelkerke pseudo R² statistics to know the proportion of variability explained by such statistical model. We used Hosmer-Lemeshow test to assess goodness of fit of the model. We also performed multiple linear regression analyses with BPRS total scores as the dependent variable, adjusted for the effects of age, gender and serum clozapine levels. We analysed our data using statistical software packages, SPSS 16.0 and STATA 12.0.

Results

Participant Characteristics

We assessed 113 consecutive patients, satisfying the eligibility criteria. We excluded six patients, who were not completely compliant with clozapine, within the past 12 weeks. One patient with severe Parkinson’s disease and another with moderate mental retardation were also excluded. Among the 105 patients, confirmed to be eligible, 101 consented to participate, making the response rate as 96.2%. Common reasons for refusing consent were lack of interest in study objectives and reluctance to provide blood samples. Participants (n = 101) and those who were excluded (n = 12) did not differ significantly on gender (χ² = 0.04; df = 1; p = 0.84), age (t = –1.41; df = 111; p = 0.16) and on their duration of illness (t = –1.27; df = 111; p = 0.21). There were 65 (64.4%) clozapine responders and 36 (35.6%) non-responders, who had BPRS total scores, 36 or more.

The majority of the sample were male (n = 73; 72.3%), unemployed (n = 60; 59.4%) and living in urban areas (n = 58; 57.4%). Their mean age was 35.43 years (SD 9.43) and their mean Body Mass Index (BMI) was 24.54 (SD 4.64). They had an average of 11.86 years of education (SD 3.89) and an average monthly family income of Indian Rupees (INR) 4733 (SD 6062) [US ≥ 106.36 (SD 136.22)]. Most of the participants had paranoid subtype (n = 85; 84.2%) of schizophrenia. Their mean duration of illness, age of onset of illness and duration of untreated psychosis were 12.40 years (SD 6.77), 23.07 years (SD 7.22) and 11.21 months (SD 13.38) respectively. Their average duration of antipsychotic drug treatment for schizophrenia was 113.64 months (SD 78.46) and their mean duration of clozapine treatment was 41.61 months (SD 39.58; median 28 months; range 4–174 months). Their mean scores on BPRS, ACE-R, WHODAS-II, CTES, RTES and PAS, were 34.73 (SD 12.45), 63.11 (SD 20.78), 17.49 (SD 12.98), 8.32 (SD 10.48), 5.94 (SD 8.71) and 54.83 (SD 21.29) respectively. Mean daily oral dose of clozapine was 340.84 mg/day (SD 119.04; median 350 mg/day; range 100–650 mg/day). Mean serum clozapine level was 550.53 ng/ml (SD 378.46; median 428 ng/ml).

Clinical Predictors of Response to Clozapine

We present the bivariate analyses for clinical predictors of response to clozapine in Table 1. Past history of catatonia, smoking more than one pack/day, excessive sedation and cognitive deficits were significantly associated with non-response to clozapine. We performed multivariate analyses adjusting for the effects of age, gender as well as serum clozapine levels. We provide those adjusted odds ratios in Table 2, which confirmed the findings of earlier bivariate analyses. A multiple logistic regression model including these four variables could explain 31.9% of variability observed in the response to clozapine (Nagelkerke R² = 0.319). Hosmer-Lemeshow test confirmed the goodness of fit of this model (χ² = 0.25; df = 5; p = 0.99). These four variables could predict the good clinical response to clozapine correctly in 95.4% of participants.

Table 1. Clinical Variables Associated with Non-response to Clozapine Among the Responders (n = 65) and Non Responders (n = 36) During Bivariate Analyses (n = 101).

	RESPONDERS	NON-RESPONDERS	BIVARIATE STATISTICS ODDS RATIO
FACTOR	n (%)	n (%)	(95% CI)	p-VALUE
Female gender	15 (23.1)	13 (36.1)	1.88 (0.77−4.60)	0.16
Never married/ divorced	39 (60.0)	27 (75.0)	2.00 (0.81−4.93)	0.13
Completed school education	57 (87.7)	33 (91.7)	1.54 (0.38−6.23)	0.54
Family history of schizophrenia	11 (16.9)	6 (16.7)	0.98 (0.33−2.92)	0.97
Paranoid subtype	56 (86.2)	29 (80.6)	0.67 (0.23−1.97)	0.46
Onset before 18 years of age	17 (26.2)	10 (27.8)	1.09 (0.44−2.71)	0.89
Past history of catatonia	0 (0)	5 (13.9)	-	0.005a
Any Axis II diagnosis	3 (4.6)	3 (8.3)	1.88 (0.36−9.83)	0.46
Duration of illness >5 years	60 (92.3)	30 (83.3)	0.42 (0.12−1.48)	0.18
DUP more than one year	17 (26.2)	12 (33.3)	1.41 (0.58−3.43)	0.45
Prior poor drug complianceb	25 (38.5)	13 (36.1)	0.90 (0.39−2.10)	0.82
Past history of ECT	24 (36.9)	13 (36.1)	0.97 (0.41−2.25)	0.94
Clozapine level <350 ng/ml	18 (27.7)	11 (30.6)	1.15 (0.47−2.81)	0.76
Clozapine duration >1 year	52 (80.0)	25 (69.4)	0.57 (0.22−1.45)	0.24
High Caffeine intakec	15 (23.1)	9 (25.0)	1.11 (0.43−2.87)	0.83
Smoking ≥ one pack/day	7 (10.8)	10 (27.8)	3.19 (1.09−9.30)	0.03
Excessive sedation on clozapined	45 (69.2)	32 (88.9)	3.56 (1.11−11.40)	0.03
Sialorrhoea on clozapine	28 (43.1)	19 (52.8)	1.48 (0.65−3.35)	0.35
AIMS	5 (7.7)	6 (16.7)	2.40 (0.68−8.51)	0.18
Childhood traumae	33 (56.8)	15 (41.7)	0.69 (0.30−1.58)	0.38
Recent Traumaf	26 (40.0)	12 (33.3)	0.75 (0.32−1.76)	0.51
Cognitive deficitsg	27 (41.5)	26 (72.2)	3.66 (1.52−8.83)	0.004
Poor premorbid adjustmenth	28 (43.1)	19 (52.8)	1.48 (0.65−3.35)	0.35

^aFisher’s exact test, Odds ratios could not be calculated;

^bAfter initiating antipsychotic therapy, had been off medication for longer than six months duration;

^cThree or more cups of coffee or tea intake/day;

^dmore than 9 hours of sleep/day;

^eChildhood Traumatic Event Scale total score >4;

^fRecent Traumatic Event Scale total score >2;

^gAddenbrooke’s Cognitive Examination—Revised total score <68;

^hPremorbid Adjustment Scale total score >54; DUP: Duration of Untreated Psychosis;

AIMS: Abnormal Involuntary Movements; ECT: Electro Convulsive Therapy.

Table 2. Multivariate Analysesa of the Clinical Predictors for Non-Response to Clozapine, Employingmultiple Outcome Definitions.

OUTCOME DEFINITION FOR CLOZAPINE NON-RESPONSE	NON-RESPONDERS n (%)	PAST HISTORY OF CATATONIAa	SMOKING ≥ ONE PACK/DAY	EXCESSIVE SEDATIONb	COGNITIVE DEFICITSc	PARANOID SUBTYPE	HIGH CAFFEINE INTAKEd	FEMALE GENDER
BPRS total score >35	36 (35.6)	p . 0.005e	5.03 (1.52−16.64) p . 0.008	4.25 (1.19−15.20) p . 0.03	3.64 (1.43−9.23) p . 0.007	NS	NS	NS
At least one of the five BPRS itemsf >3	30 (29.7)	18.16 (1.81−182.38) p = 0.01	NS	NS	NS	NS	NS	NS
BPRS total score ≥38 (worst quartile)	25 (24.8)	p . 0.001e	NS	NS	3.71 (1.28−10.79) p = 0.02	0.27 (0.08−0.87) p = 0.03	NS	NS
At least two of the five BPRS itemsf >3	15 (14.9)	46.34 (4.51−476.66) p = 0.001	15.61 (2.24−108.70) p = 0.006	NS	NS	NS	7.28 (1.77−29.98) p = 0.006	4.54g (1.30−15.84) p = 0.02

NS: statistically not significant (p value ≥ 0.05);

^aMultiple logistic regression models to calculate Adjusted Odds Ratios (95% CI), accounting for the effects of age, gender and serum clozapine levels (ng/ml);

^bmore than 9 hours of sleep/day;

^cAddenbrooke’s Cognitive Examination—Revised total score <68;

^dThree or more cups of coffee or tea intake/day;

^eFisher’s exact test two tailed p value;

^fSuspiciousness, hallucinatory behaviours, grandiosity, conceptual disorganization and unusual thought content;

^gAdjusted for the effects of age and serum clozapine levels (ng/ml).

Secondary Analyses

We repeated similar analyses using three more outcome definitions for non-response to clozapine. We present the results of those multivariate analyses in Table 2. The clinical predictors varied, when different outcome definitions were employed. However, past history of catatonia, defined by one or more catatonic symptoms listed in DSM IV-TR, remained as a consistent clinical predictor of non-response, while using all definitions. In order to dispel the concerns over the dichotomization of response to clozapine, we performed multiple linear regression analyses with BPRS total scores as the dependent variable, adjusted for the effects of age, gender and serum clozapine levels. Past history of catatonia (β = 22.75; SE = 5.16; t = 4.41; p = 0.001), smoking more than one pack/day (β = 5.40; SE = 2.64; t = 2.05; p = 0.04) and cognitive deficits (β = 7.73; SE = 2.41; t = 3.20; p = 0.002) were significantly associated with higher BPRS total scores. The clozapine responders and non-responders, who had BPRS total scores, 36 or more, did not differ on their prescription for any psychotropic co-medication (χ² = 0.34; df = 1; p = 0.56).

Correlates of Adverse Events

Our participants had the following adverse events, related to clozapine, sialorrhoea (n = 47; 46.5%), nausea or vomiting (n = 21; 20.8%), constipation (n = 21; 20.8%), erectile dysfunction (13 men; 27.7%), dyslipidemia (n = 12; 11.9%), seizures (n = 9; 8.9%) and nocturnal enuresis (n = 6; 5.9%). Fifteen participants (14.9%) were obese (BMI ≥ 30) and 23 (22.8%) were overweight (25 ≤ BMI < 30). As we never re-challenge patients, who have developed neutropenia on clozapine, none of our participants had past history of neutropenia or agranulocytosis. The following variables were significantly associated with each other, after adjusting for the effects of serum clozapine levels, by multiple logistic regression: age above 42 years and clozapine related seizures (AOR 6.37; 95% CI 1.32–30.78); worst quartile BPRS total scores (≥38) and nocturnal enuresis (AOR 7.24; 95% CI 1.22–42.87); poor premorbid adjustment scores and obesity (AOR 3.80; 95% CI 1.11–12.94).

Correlates of Disability

BPRS total scores (β = 0.22; SE = 0.10; t = 2.09; p = 0.04) had significant linear relationship with WHODAS-II 12 item total disability scores. The median value of WHODAS-II total scores of our participants was 15 (range 0–48). Participants, who scored more than that threshold, were categorized to have higher disability. Female gender (AOR 2.69; 95% CI 1.08–6.70) and cognitive deficits (AOR 3.41; 95% CI 1.44–8.10) were significantly associated with higher disability after adjusting for the effects of age and serum clozapine levels.

Discussion

This study investigated the clinical predictors of response to clozapine among patients with TRS. Our results suggest that past history of catatonia, smoking more than one pack/day, hyper-somnolence and cognitive deficits were associated with non-response to clozapine. We documented that the clinical predictors varied, depending on the outcome definitions for non-response to clozapine.

Strengths and Limitations

The strengths of this study include, exclusively recruiting patients with TRS, relatively larger^14-19 sample size, minimal refusal rate, estimation of serum clozapine levels and employing standard instruments to assess premorbid adjustment, traumatic life events, cognition and disability. Consecutive sampling strategy reduced the possibility of selection bias. The independent assessment of exposure and outcome variables by two different investigators minimized potential observer bias. Interviewing primary caregivers of the participants and verifying their follow-up medical records reduced the possibility of recall bias on the reported clinical variables. Clinical predictors of response to clozapine often overlap with clinical predictors of poor prognosis in schizophrenia.⁶ Exclusively recruiting patients with TRS made our results more specific.

The potential limitations of this study include the cross-sectional clinical assessment of response to clozapine and the temporal ambiguity. Our participants were on clozapine for a mean duration of 41.61 months. We recruited only the participants, who were maintained on stable dosage of clozapine for a minimum duration 12 weeks, when their treating psychiatrists did not need to change their prescription. Hence, their cross-sectional BPRS scores were more indicative of the persistent psychopathology than of any acute fluctuations in their illnesses. As patients with TRS may respond to clozapine after a delay of 24–32 weeks,^15,33 short-term longitudinal studies, with less than 12 weeks of follow up, also had similar temporal uncertainties.^14,18 Considering the pragmatic constraints and the paucity of pertinent data in resource-poor settings, we determined the response to clozapine among our participants with the cross-sectional BPRS assessment, employing multiple outcome definitions.

The dichotomous categorization of clinical response to clozapine is debatable. Despite many researchers define the response to clozapine by the reduction in the total scores of BPRS, most clinical psychiatrists prefer to use the discrete clinical category of non-response based on the presence of persistent positive or negative symptoms.³¹ Clinical significance of many statistically significant reductions in the total scores of rating scales is uncertain.³⁴ Hence, we analyzed multiple categorical outcome definitions and BPRS total scores, as dependent variables, by appropriate multiple logistic regression and multiple linear regression models to confirm our findings. Despite the extensive use of BPRS, we should acknowledge that there are more diverse outcome measures to assess the treatment responses in schizophrenia.³⁵

Disparities among Clinical Predictors

Female gender,¹⁷ earlier age of onset,^17,21 non-paranoid subtype,³⁶ longer duration of illness,¹¹ baseline psychopathology,^14,16,37 baseline quality of life,³⁷ serum clozapine levels below 350 ng/ml¹¹ and poor functioning during the previous year^6,15 have already been reported as the potential clinical predictors of non-response to clozapine. Findings on the utility of past history of neuroleptic induced extra pyramidal symptoms^11,16,38 and of baseline negative symptoms^16,21 as clinical predictors were contradictory. Ethnic diversity, population characteristics, lack of structured assessment of clinical variables, differing outcome definitions and causal heterogeneity of response to clozapine may explain the disparities between these findings. Our study was negative for these potential clinical predictors, but it has brought forth four clinical variables, which could explain 31.9% of variability in the treatment response to clozapine. When we employed alternative outcome definitions, female gender, high caffeine intake and non-paranoid subtype were also associated with non-response to clozapine. Our results add evidence that differing outcome definitions can, at least partly, explain the disparities among the previous studies on this topic.

Schizophrenia is not a single disease, but a heterogeneous disorder, caused by multiple genetic as well as environmental factors.³⁹ Pharmacokinetics and pharmacodynamics of clozapine is complex.⁴⁰ Hence, any statistical model with limited clinical or biological predictors often explains only a small proportion of variability among the responses to clozapine in TRS. Differing BPRS based response cut-off thresholds are known to cause significant variability in the results of antipsychotic drug trials.⁴¹ Though a primary BPRS cut-off score, based on clinical relevance, should be chosen a priori, further analyses with wider range of cut-off scores should be presented to explain more variability in the treatment response.⁴¹ Hence, our results support the need for employing multiple outcome definitions to address the heterogeneity of TRS and of its response to clozapine.

Catatonia and Response to Clozapine

Debate on the nosological status of catatonia, as a sub-type of schizophrenia or as a separate syndrome, remains unresolved. The patients who have both schizophrenia and catatonia may have unique underlying pathophysiology.⁴² Reports on catatonia, increasing the risk for clozapine induced neuroleptic malignant syndrome,⁴³ and on clozapine withdrawal catatonia⁴⁴ are available. Unlike the earlier case reports, stating good response to clozapine in patients with recurrent catatonia,^45,46 all participants with past history of catatonia (n = 5) were clozapine non-responders in this study. Investigating the relationship between catatonia and its response to clozapine as well as the possible underlying neurobiological processes are warranted.

Clinical Recommendations for Modifiable Risk Ffactors

Smoking reduces serum clozapine levels.⁴⁷ However, available literature on the relationship between smoking and the response to clozapine are contradictory.^48,49 Studies associating smoking with non-response to clozapine emphasize its ability to induce Cytochrome P-450 1 A2 (CYP1 A2) enzyme and to reduce serum clozapine levels.⁴⁸ Studies associating smoking with good response to clozapine focus on its ability to activate nicotinic receptors.⁴⁹ Our results have documented that, smoking more than 20 cigarettes/day, increased the risk of non-response to clozapine by five times, after adjusting for the effects of age, gender and serum clozapine levels. We infer that the relationship between smoking and non-response to clozapine cannot be solely explained by reduced serum clozapine levels. We highlight smoking as a potentially modifiable risk factor and suggest nicotine deaddiction for all smoking patients with TRS to augment their clozapine response. As concerns over abrupt smoking cessation, which may suddenly elevate the serum clozapine levels to toxic range, exist, we suggest appropriate titration of clozapine dosage against its serum levels during smoking cessation.⁵⁰

Daily consumption of caffeinated beverages is common globally. Our findings suggest the need for motivating all patients on clozapine, especially those who have poor response to clozapine, to reduce their caffeine consumption, for augmenting their clinical response.⁴⁸ Akin to previous studies reporting the ability of cognitive factors discriminating clozapine responders and non-responders,⁵¹ we found cognitive deficits as an important predictor for non-response. We also documented that cognitive dysfunction was associated with higher disability. We suggest augmenting the clinical response to clozapine with appropriate cognitive interventions, for such disabled patients. Despite temporal ambiguity, hyper-somnolence was associated with non-response to clozapine. Complaints of excessive sedation on clozapine are often overlooked during clinical practice. Appropriate behavioural management and adjuvant psychopharmacological agents to reduce excessive sedation should be considered.

Conclusions

The inherent heterogeneity within the clinical category of TRS obscures the search for the clinical and biological predictors for its response to clozapine. Syndrome sub-categorization employing biological variables is desired to achieve the etiological homogeneity of schizophrenia.⁵² Appropriate biological parameters to categorize response to clozapine have to be identified for developing better prediction models with clinical or biological variables. As explanatory pluralism is the need of the hour,⁵³ it is high time to reduce the wide gulf between the clinical and pharmacogenetic research on this topic. Combining clinical and pharmacogenetic predictors may identify more patients, who are most likely to benefit with clozapine, and may prevent the unnecessary exposure of many potential non-responders to serious adverse effects. We suggest future longitudinal studies, investigating both clinical and pharmacogenetic factors together with multiple outcome definitions, to predict response to clozapine in patients with TRS.