Factors Associated with Response to Clozapine in Schizophrenia: A Review

Abstract

Background

Clozapine has been serving as the gold standard medication for patients with treatment-resistant schizophrenia who failed to respond to other antipsychotics. However, factors affecting response to this medication have not been comprehensively reviewed recently. Methods: In order to find factors associated with response to clozapine in schizophrenia, a literature search was conducted using PubMed through January 2011 with keywords of clozapine, response, and schizophrenia. Cross-referencing of relevant articles was also performed. Factors were arbitrarily classified into the following: demographic/clinical, oral dosage/pharmacokinetic, biochemical, (electro)physiological, genetic, imaging, and combinations.

Results

A synthesis from 280 articles indicated that demographic and clinical variables such as high baseline symptoms and low premorbid functioning have not been particularly useful in predicting response to clozapine. Pharmacokinetic evidence points to a threshold clozapine level of 350 ng/ml but in a context of significant inter- as well as intra-individual variability. Pharmacokinetic perspectives appear to have more implication in special situations including poor response, suspected toxicity and nonadherence. A number of laboratory-based studies have reported on many potential candidates for response prediction to clozapine, however, reproducibility, specificity, robustness of the findings, as well as clinical feasibility and cost-effectiveness all pose a significant practical challenge, in relation with the fact that pathophysiological bases of treatment resistance in schizophrenia largely remain to be elucidated.

Conclusions

No unequivocal factors to clozapine response were found despite a relatively rich body of the literature, which calls for more works on this important topic. Clozapine level of 350 ng/ml appears to be useful in case of nonresponse.

Introduction

It has been known that a clinical course of schizophrenia is heterogeneous with some patients showing progressively deteriorating outcome.¹ It is estimated that roughly a third of patients are classified as treatment-resistant schizophrenia.² Treatment resistance may be underlined with unique pathophysiological processes³ but it remains largely unknown as to whether these reported changes represent totally independent pathophysiological process unique to resistant patients, more pronounced process in schizophrenia in general, or rather accelerated physiological process.⁴ Nor are they not always easy to disentangle as state or trait markers. For instance, while morphological variables (i.e., cortical loss and ventricular enlargement) appear to characterize this population, functional significance of morphological alterations in relation to treatment response has not necessarily been clear,⁵ in addition to an absence of neurodegeneration or ongoing neuronal injury in a postmortem study of the elderly patients⁶ who represent chronicity of the illness. As such, it is implied that morphological imaging as a screening tool adds little to actual management of the illness with cost-effectiveness remaining an equivocal issue.⁷

This noted, since the pivotal study by Kane et al.⁸ found that about a third of stringently defined patients with treatment-resistant schizophrenia responded to clozapine, it has been the gold standard medication in this specific population. In fact, subsequent clinical trials have been consistent (albeit with some nuance) to show its superiority over others,^9–12 and a meta-analysis confirmed the largest effect size among the newer generation antipsychotics.¹³ Moreover, clozapine was associated with a lower rate of discontinuation for any reason¹⁴ and a substantially lower mortality in schizophrenia.¹⁵ It is also important to point out that patients in research settings might be less ill,¹⁶ which nevertheless does not argue for a less room for improvement with clozapine among real-world patients.¹⁷ However, factors affecting response to this medication have not been comprehensively reviewed recently. This paper addressed this highly pertinent issue.

Methods

A review of the literature was conducted to find factors associated with response to clozapine using PubMed (last search 31 January 2011). Keywords were ‘clozapine’, ‘response’, and ‘schizophrenia’, and limits were set with ‘Humans’ and ‘English’. This yielded 825 hits. Cross-referencing of the relevant articles was also performed. In this paper, factors were arbitrarily classified into the following: clinical/demographic, oral dosage/pharmacokinetic, biochemical, (electro)physiological, genetic, imaging, and combinations. Where applicable, factors reported for illness chronicity and treatment resistance in schizophrenia in general were briefly discussed before exclusively focusing on factors related with clozapine response.

Results

Demographic and Clinical Factors

African American patients reportedly showed greater symptomatic improvement but they were more likely to discontinue clozapine during hospitalization compared with white patients.¹⁸ Another study¹⁹ also found that African American patients were more likely to discontinue clozapine than non-African American patients and reported a trend for a prior history of inadequate response to typical antipsychotics to predict discontinuation of clozapine, suggesting a possibility that more unresponsive patients may not do well even with clozapine. Male gender was associated with less favorable outcome in some^20,21 but not all studies.^22–24 Earlier age at illness onset was related with poorer response^22,23,25 although another study found the opposite.²⁰ A negative association between response rate and age²⁶ or longer duration of the illness²² is also reported. Patients aged 55 to 64 may respond better than those aged 65 and older²⁷ although few studies specifically targeted geriatric population.²⁸

A presence of extrapyramidal symptoms (EPS) with antipsychotic pretreatment was identified as a favorable factor^22,29 although this was not a case in another study.³⁰ A significance of this old finding is difficult to interpret in light of a recent predominance of atypical agents that are associated with relatively less liability for EPS in general.¹³ Smokers reportedly showed significantly greater therapeutic response to clozapine than nonsmokers despite they smoked less when treated with clozapine than when treated with conventional antipsychotics.³¹ It is reported that clozapine-induced weight gain predicted improvement in psychopathology³² and it was associated with the baseline body mass index as well as clozapine dosage.³³ An initial response to clozapine was reportedly associated with subsequent long-term weight gain.³⁴ However, a specificity of this finding seems questionable in that a similar situation was observed with olanzapine for instance,^35,36 and that long-term changes in the body mass index were found to be similar between clozapine and typical antipsychotics in another study.³⁷ Increase in lipid levels is implicated for favorable response to clozapine, potentially serving as a reservoir independently of weight.³⁸

It was found that schizoaffective disorder and bipolar disorder patients were more responsive to clozapine than schizophrenia patients.^20,39,40 It is reported that behavioral disorganization was higher in partial responders while cognitive disorganization was higher in nonresponders.⁴¹ Comorbid obsessive-compulsive disorder⁴² as well as akathisia⁴³ appears to complicate treatment while substance abuse might not.⁴⁴ Paranoid patients^22,23 in contrast with deficit patients⁴⁵ appear to be more responsive. Poor functioning in the previous year⁴⁶ or an absence of independent living history⁴⁵ was associated with less favorable response. Low premorbid social functioning and individual autonomy were a negative predictor but intellectual functioning was not,⁴⁷ while those with higher education were more likely to achieve functional outcome in another study.²⁰ It is also indicated that clozapine nonresponders tended to be less social and more withdrawn than responders.⁴⁸

Four factors of negative symptoms, reality distortion, disorganization, and anxiety/depression are proposed to predict response to antipsychotics including clozapine.⁴⁹ Response was reportedly more likely when negative symptoms were severe²⁵ and clozapine may have some direct effect on core negative symptoms.⁵⁰ However, it may also be true that clozapine is of limited value for negative symptoms that are presumed to be primary^51,52 and clozapine may not show a distinct benefit for negative symptoms or deficit syndrome.⁵³ Honigfeld and Patin⁵⁴ failed to find any clinical or demographic predictors. Rosenheck et al.⁵⁵ did not find specific predictors of improvement other than high baseline symptoms in their systematic large study comparing clozapine and haloperidol.⁹ A more symptom severity as a predictor of more robust response was also noted in another study.⁵⁶ High scores in symptoms together with less severity and less negative symptoms, are related to good response in a sample of ambulatory patients,³⁰ which appears to be all plausible but nonspecific.

Oral Dosage and Pharmacokinetic Factors

In the pivotal studies, clozapine was dosed at up to 900 mg (mean maximum dose of about 600 mg)⁸ and 100–900 mg (mean 552 ± 229 mg),⁹ while the dose was more modest (at about 333 mg/day) in more recent trials.^11,12 There is a suggestion that many patients require a dosage of 400 mg or more⁵⁷ and clozapine shows a dose-related efficacy up to 600 mg according to limited data.⁵⁸ One study indicated a significant dose effect (600 > 300 > 100 mg/day) at 16 weeks of treatment.⁵⁹ In contrast, a retrospective study was suggestive of a usefulness for lower (294 mg) compared with higher dosage (525 mg).⁶⁰ Furthermore, an interesting observation noted that clinical efficacy of a lower dosing regimen in Austria was somewhat paradoxically superior to a higher one in the U.S.⁶¹ with less adverse effects (284 mg versus 444 mg).⁶²

A number of studies related blood level with clinical response, implicating a clozapine concentration of more than 350 ng/ml is necessary for response.^22,63–67 Depending on a population studied, a higher threshold of 420 ng/ml^67,68 or 550 ng/ml is also reported.⁶⁹ Furthermore, plasma levels of as high as 1077 ± 457 ng/ml were reported although symptomatology and functional disability were similar in those whose levels were over versus under 1000 ng/ml (at oral dosage of 900 mg).⁷⁰ Another study showed a superior efficacy of 200–300 ng/ml and 350–450 ng/ml over 50–150 ng/ml range with no advantage for the higher over the middle.⁷¹ Kurz et al.⁷² reported a lower threshold for improvement (132 ± 84 ng/ml with oral dosage of 225 ± 117 mg). Xiang et al.⁷³ reported that prevention of relapse may require maintenance of patients at concentrations above 200 ng/ml and above 60% of the acute-phase level. On the other hand, a plasma clozapine concentration of over 250 ng/ml was reportedly more associated with moderate/severe side effects.⁷⁴ According to Spina et al.,⁷⁵ the incidence of side effects was higher at clozapine concentrations above 350 ng/ml compared with lower ones (38% versus 17%).

Nonresponders reportedly exhibited mean clozapine plasma levels constantly below the value of 260 ng/ml with N-demethylation being the preferred metabolic route, while early responders showed higher levels and late responders a progressive increase in levels to a plateau despite a fixed dosage.⁷⁶ However, Mauri et al.⁷⁷ found a higher mean metabolic ratio (N-desmethylated metabolite norclozapine/clozapine) in responders, and this finding was supported in a separate sample of adult patients⁷⁸ as well as in younger patients with childhood-onset schizophrenia.⁷⁹ Younger patients may more avidly produce norclozapine.⁸⁰ Perry et al.⁶³ suggested a threshold of 450 ng/ml for clozapine plus norclozapine. Norclozapine may be important in mediating antipsychotic effect,⁸¹ where muscarinic M1 receptor agonism⁸² and a resultant potentiation of hippocampal N-methyl-D-aspartate (NMDA) receptor currents⁸³ and increased release of acetylcholine and dopamine⁸⁴ may be relevant. A high clozapine/norclozapine ratio is implicated in cognitive impairment in schizophrenia.^82,85 Blood levels appear to be more precisely interpreted in the context of cytochrome P450 isozyme (CYP) 1A2 genotype, which is an important determinant for dosage as well as clearance.⁸⁶ It is the key enzyme involved in converting clozapine to norclozapine while CYP3A4 is the primary enzyme involved in conversion to clozapine- N-oxide,⁸⁷ although other genotypes appear to be involved in drug interactions of clozapine.^88,89 CYP1A2 activity may simply be phenotyped with caffeine metabolic ratio in overnight urine.⁹⁰

It is reported that clozapine and norclozapine levels were significantly higher in patients with obsessive compulsive symptoms than those without (595 ± 365 and 266 ± 144 ng/mL versus 434 ± 253 and 203 ± 120 ng/mL, respectively).⁹¹ Lane et al.⁹² found that each 1-year increment in age elevated clozapine level by 1.1%, norclozapine by 1.0%, and clozapine-N-oxide by 1.0%. Smoking and male gender were associated with lower exposure to clozapine and norclozapine due to the higher oral clearance.⁹³ Oyewumi et al.⁹⁴ found that trough drug levels at the start of clozapine may predict individualized target dose at steady state. Perry et al.⁹⁵ proposed an equation to predict steady-state concentration and a dosing nomogram. Rostami-Hodjegan et al.⁹⁶ suggested another by incorporating dose, plasma norclozapine/clozapine ratio, sex, age weight, and smoking, thus accounting for 48% of the observed variation in plasma clozapine concentration (ranging from 10 to 350 ng/mL). Paz et al.⁹⁷ evaluated these three methods, and found that the predictive model of Oyewumi et al.⁹⁴ was of clinical value for predicting clozapine concentration or the dose required to achieve a specified concentration.

Important to note, however, is significant inter-individual variability. If fact, an up to an eightfold variability at a given dose is reported in inpatients⁹⁸ for whom adherence would be less of a concern. Further, Potkin et al.⁶⁸ noted as high as 45-fold variability (40–1911 ng/mL) with a fixed-dose treatment at 400 mg/day. A Chinese sample reported a high level of 1078 ± 385 ng/ml with a moderate oral dose of 373 ± 90 mg/day.⁹⁹ The mean ± SD serum concentrations of clozapine and norclozapine among Korean patients were 611 ± 368 ng/ml and 315 ± 163 ng/ml, respectively at 323 ± 125 mg/day.⁷⁸ It was found Asian versus Caucasian patients had more than twice the effective clozapine concentration to dose ratio.¹⁰⁰ Intra-individual variability affected by simple daily factors is also substantial, rendering blood level less stable and unreliable marker. For instance, smoking cessation was found to result in a significant increase in the number of patients showing blood level of 1000 ng/ml or more from 4.2% to 41.7%.¹⁰¹ In contrast, a daily consumption of 7–12 cigarettes may be sufficient for maximum induction of clozapine metabolism, suggesting a need for a 50% lower starting dose among nonsomkers.¹⁰² Also, abstinence from caffeine is reported to decrease clozapine level by 47%.¹⁰³

Biochemical Factors

Treatment resistance in schizophrenia may be associated with biochemical alterations such as increased interleukin (IL)-8 and IL-10,¹⁰⁴ IL-6,¹⁰⁵ and IL-1 receptor antagonist levels.¹⁰⁶ An increase in cortisol together with IL-2 and IL-6 is reported as well.¹⁰⁷ Chronically institutionalized patients reportedly showed lower plasma brain-derived neurotrophic factor (BDNF) levels.^108,109An increase in S100B (a family of calcium-binding proteins) was found in an older patient population.¹¹⁰ A postmortem study showed increased glycine in the orbitofrontal cortex and reduced serine in the putamen as well as threonine in the cerebral cortices among chronic patients.¹¹¹ An increase of strychnine-insensitive glycine binding sites in several cerebral areas is reported,¹¹² and another study found reduced tyrosine hydroxylase-labeled axodendritic synapses in resistant patients, implicating a lower density of dopaminergic synapses.¹¹³ Patients most likely to continue to manifest symptomatology in late-life may be those with low platelet monoamine oxidase (MAO) activity.¹¹⁴ Nonresponders to antipsychotics may harbor lower cerebrospinal fluid (CSF) or plasma catecholamine levels.¹¹⁵

Clozapine is known to suppress thymidine incorporation by lymphocytes and secretion of IL-2, IL-4 and interferon-gamma, although such effects were nonspecific in that they were observed with haloperidol as well.¹¹⁶ Clozapine reportedly increased plasma IL-6 and CC16 (Clara cell protein, an endogenous anti-cytokine) earlier (in the first 2 weeks of treatment), whereas increases in plasma soluble CD8 antigen and IL-1 receptor antagonist were noted later (after 5 weeks).¹¹⁷ Clozapine treatment was reported to be associated with a rise in high sensitive C-reactive protein,¹¹⁸ and a higher serum clozapine level with increased antiphospholipid antibodies.¹¹⁹ Higher glycine levels and glycine/serine ratios at baseline were reportedly associated with greater improvements in negative symptoms.¹²⁰ Reduction in serum glutamate levels is also implicated in clozapine treatment.¹²¹ Greater decrease in prolactin (PRL) and increase in growth hormone (GH) response to apomorphine stimulation pretreatment, and greater inhibition of PRL and GH response to apomorphine posttreatment, were biochemical characteristics pointed out by Lieberman et al.²² Improvement was found to correlate significantly with reduction in PRL response to d-fenfluramine challenge.¹²² It is reported that clozapine responders showed significantly higher adrenocorticotropic hormone (ACTH) responses to m-chlorophenylpiperazine (a proxy of serotonergic function) during drug-free state than nonresponders.^123,124 Joffe et al.¹²⁵ indicated that clozapine-induced decrease in production of reactive oxygen metabolites by monocytes in vitro may predict clinical response.

Pickar et al.¹²⁶ found low ratio of pretreatment homovanillic acid (HVA)/5-hydroxyindoleacetic acid (5-HIAA) in CSF as a predictor of clozapine response. Lower epinephrine levels pretreatment and increased mean plasma concentrations of methoxyhydroxyphenylglycol (MHPG) and epinephrine during treatment were implicated in early-onset cases.^127,128 Plasma levels of HVA and MHPG were reported to decrease during the initial weeks of treatment in clozapine responders.¹²⁹ Pretreatment HVA/5HIAA ratio (and HVA level to a lesser extent) in CSF reportedly predicted treatment response in another study.¹³⁰ Dynamic temporal changes may be important such as an initial increase at 4 weeks then a decrease in serotonin levels.¹³¹ Elevation in norepinephrine levels induced by clozapine may result from enhanced neurotransmitter spillover,¹³² and may provide a basis for antiaggressive and antisuicidal effects¹³³ or improvement in negative but not positive symptoms.¹³⁴ It is reported that low baseline plasma serotonin and platelet MAO levels were increased by clozapine treatment, which explained a minority of improvement.¹³⁵ However, monoamine levels failed to show significant changes during clozapine treatment in the study by Jacobsen et al.¹³⁶ who exclusively studied patients with the onset of illness by age 12.

(Electro) Physiological Factors

Peripheral indicators of autonomic activity indicated that clozapine markedly attenuated electrodermal base levels and both phasic and tonic electrodermal responsivity compared to placebo (and somewhat less consistently compared to fluphenazine), and that a smaller autonomic response to the mild stress of task performance and larger heart rate responses to nonsignal tones on the alternate treatments were predictive of a good clinical response to clozapine.¹³⁷ A reduction in skin conductance rate was reported in comparison with olanzapine.¹³⁸ Patients taking clozapine were found to show significantly higher heart rate, lower heart rate variability and lower high-frequency and higher low-frequency components¹³⁹ and evidence of cardiovascular autonomic neuropathy.¹⁴⁰ It is reported that body temperature was inversely related to clozapine dose and higher nor-clozapine/clozapine ratio was associated with higher blood pressure, although the magnitude of associations was weak while significant.¹⁴¹ Clozapine is known to reduce parasympathetic resting tone,¹⁴² but a decrease of sympathetic activity in the autonomic nervous system might present as greater heart rate variability among responders.¹⁴³ A significant correlation was reported between patients’ psychopathology scores and their mean weekly corneal temperature, although there was no significant difference in patients and controls.¹⁴⁴

High amplitude of intrusive saccades in drug-free patients reportedly predicted poor response to clozapine in a smooth pursuit eye movement paradigm.¹⁴⁵ Treatment-refractory patients showed pre-attentive abnormalities in auditory processing, characterized with mismatch negativity (MMN) amplitude that was not reduced and was the largest at the central-posterior location, and they displayed larger negativities at the left frontal and the left and right temporal electrodes.¹⁴⁶ Resistant patients were reported to show lower P300 amplitudes in event related potential (ERP).¹⁴⁷ It was found that clozapine treatment was associated with normalization of prepulse inhibition (PPI),¹⁴⁸ increase in P3b (active auditory information processing) in the frontal electrode,¹⁴⁹ and increase in P300 amplitude without affecting N200 and MMN.¹⁵⁰ Clozapine responders were reportedly distinguished by having a relatively intact MMN, while poor responders had significantly larger P3a amplitudes (response to unexpected events) in the ignore auditory condition.¹⁵¹

It is reported that computer analyzed electroencephalography (EEG) of resistant patients was characterized by a large amount of alpha activity and less fast activity.¹⁵² EEG abnormalities associated with clozapine appear prevalent amongst antipsychotics¹⁵³ and abnormality in EEG after clozapine might be associated with response,^154,155 while it is also indicated that EEG abnormalities even before clozapine treatment might predict response in female patients.¹⁵⁶ Patients who responded to clozapine reportedly had higher amplitudes in the alpha spectrum in computerized EEG that were most pronounced in the left anterior quadrant.¹⁵⁷ It was found that an increase of EEG photic driving correlated with clinical improvement with clozapine.¹⁵⁸ A shift in the driving system to the frontal channels was reportedly related to response in the mutual cross-prediction algorithm.¹⁵⁹ EEG slowing as shown by decreases in relative alpha power, mean beta/total spectrum frequency, and widespread increases in absolute total and delta/theta power, may be suggestive of response.¹⁶⁰ Changes of coherence but not amplitude are implicated¹⁶¹ where posttreatment intrahemispheric asymmetry may be related to response.^162,163 Recent data indicated that, by applying advanced machine learning methodologies, pretreatment EEG may serve as a possible predictor to clozapine response.¹⁶⁴

Genetic Factors

Genotype expressing low CYP 3A5 as well as the haplotype T/A/G/ A/C in dopamine D3 receptors reportedly showed an association with refractoriness to typical antipsychotics in a European-Brazilian sample.¹⁶⁵ The odds of having genotypes of both low activity in catechol-O-methyltransferase (COMT) and high activity in angiotensin-converting enzyme (ACE) were reported to be over 10 times higher among nonresponders to typical antipsychotics.¹⁶⁶ Antipsychotic maintenance dose was found to be higher in those with L/L COMT genotype than others.¹⁶⁷ It is reported that patients carrying both NOTCH4 C/C genotype and COMT low/low genotype had more than ten times higher risk of being nonresponders to typical antipsychotics.¹⁶⁸ The MnlI polymorphism in synaptosomal-associated protein of 25 kDa (SNAP-25) was found to be associated with response to antipsychotics.¹⁶⁹ A significant amount of telomere shortening in leukocytes from poorly responsive patients is reported.¹⁷⁰

Having one or more serotonin 5-HT2C receptor Ser alleles (Cys23Ser) may predict good response to clozapine¹⁷¹ but another study failed to replicate the finding.¹⁷² A/C/C haplotype composed by rs2276302/rs1062613/rs1150226 in 5-HT3A receptor was associated with better response¹⁷³ and antipsychitic maintenance dose for those with T/T genotype at rs10622613 was significantly higher in another study.¹⁷⁴ It was found that homozygosity for C102 allele (T/C) in the coding region of 5-HT2A gene was more frequent among nonresponders^175,176 but other reports failed to replicate such an association,^177–180 a similar contradictory situation of which was also a case for 452-His/Tyr polymorphism in this gene.^{177,179,181,182} Homozygosity for allele G-1438 (G/A) in the promoter region 5-HT2A gene was reportedly higher in nonresponders,¹⁸² which was not observed in another study.¹⁷⁹ Low expression allele S´ in serotonin transporter gene is recently implicated in clozapine nonresponse.¹⁸³

Those with allele A (308G/A) of the tumor necrosis factor (TNF), which is also implicated in weight gain,¹⁸⁴ may respond better to clozapine¹⁸⁵ but inconsistently.¹⁸⁶ It is reported that homozygosity for the T825 allele (C/T) in G-protein GNB3 gene was more frequent among nonresponders,¹⁸⁷ an effect of which may be attributable to Caucasians.¹⁸⁸ Those with diplotype ACCCTC/GTTGCC, genotypes T/T + T/C, or allele T of marker rs742105 (P1333) in the dysbindin DTNBP1 gene reportedly showed better response to clozapine.¹⁸⁹ Brain-derived neurotrophic factor (BDNF) Val66Met polymorphism is implicated in response,¹⁹⁰ and T/G/G haplotype composed by rs1128397-rs13250096-rs4567028 in glial-cell-line derived neurotrophic factor (GDNF) family receptor alpha 2 is reported to be associated with positive response to clozapine.¹⁹¹ The rs2740204 variant in oxytocin gene and rs11706648, rs4686301 and rs237899 in oxytocin receptor may be associated with treatment response and improvement in positive symptoms.¹⁹² It was found that human leukocyte antigens HLA-A1 may predict a good therapeutic outcome¹⁹³ as well as a low risk of agranulocytosis.¹⁹⁴

Treatment response may be associated with Ser-9-Gly in polymorphism in dopamine D3 receptor gene¹⁹⁵ and such as association was reported in a combined analysis,¹⁹⁶ although this was not necessary consistent in the individual study¹⁹⁷ and a specificity to clozapine is questionable.¹⁹⁸ It is reported that the polymorphic marker rs2975226 (71T/A) of the dopamine transporter (SLC6A3) gene and the haplotype block constructed from the first 3 of 6 markers near the 5´-region showed significant association with clozapine response.¹⁹⁹ It is indicated that frequencies of five 48 bp repeats homozygous genotype and five 48 bp repeats allele in dopamine D4 receptor gene were significantly less among responders.²⁰⁰ On the other hand, while significant alterations (mostly decrease) in transcripts are reported in schizophrenia,²⁰¹ they have not been adequately investigated in terms of treatment resistance or response to antipsychotics including clozapine. In addition, while dopaminergic and serotonergic pathways appear to have been a main focus,²⁰² the results regarding the most straightforward dopamine D2 receptors, on which all antipsychotics in the market act, have been complicated to interpret.^203–205

Imaging Factors

It is suggested that the right middle frontal cortex may be particularly vulnerable to chronicity of the illness.²⁰⁶ Treatment resistance has been linked to a tendency for generalized cerebral atrophy²⁰⁷ as well as ventricular enlargement and a pattern of posteriorization of gray matter deficits that uniquely involves the occipital cortex.²⁰⁸ (Parts of) the corpus callosum may be involved in treatment resistance.²⁰⁹ Schizophrenia seems to be characterized with smaller gray matter volumes in a variety of areas including the cingulate gyrus by the time of illness onset,²¹⁰ which may be then progressive.²¹¹ Furthermore, prospective brain tissue loss appears to be more prominent in poor outcome patients.^147,212 On the other hand, volume reduction in the parahippocampal gyrus may not be affected by chronicity.²¹³ It is indicated that typical antipsychotics led to increased gray matter volume of the basal ganglia while atypical antipsychotics reversed this effect after switching.²¹⁴ It is reported that cumulative exposure correlated positively with gray matter volumes in the cingulate gyrus for typical and in the thalamus for atypical antipsychotics.²¹⁵ Clozapine reportedly increased frontal gray matter volume.²¹⁶

Functional imaging using magnetic resonance spectroscopy (MRS) showed an increased level of phosphodiesters in the bilateral medial temporal lobes and a decreased level of beta adenosine triphosphate in the left medial temporal lobe.²¹⁷ Using 133Xenon inhalation technique, Mathew and Wilson²¹⁸ found a low anteroposterior gradient of cerebral blood flow among chronic patients. With a usage of single photon emission tomography (SPECT), Molina Rodríguez et al.²¹⁹ found that refractory patients showed bilateral (predominantly left-sided) hypofrontality and hypotemporality as well as increased perfusion in the right basal ganglia, although Lawrie et al.²²⁰ found a limited yield with this device. Andreasen et al.²²¹ found a lack of activation in the left mesial frontal cortex and the right parietal cortex during the Tower of London task, an effect of which was unrelated to illness chronicity but related to negative symptoms. Studies with positron emission tomography (PET) found that chronically ill patients showed lower flow in prefrontal and higher flow in thalamic and cerebellar regions,²²² and that insular hypometabolism and a certain degree of hypofrontality were associated with chroniciy of the illness.²²³

Prefrontal sulcal prominence as assessed with computed tomography (CT) is reported to be inversely related to clozapine response.²²⁴ Using magnetic resonance imaging (MRI), Lauriello et al.²²⁵ found that patients with larger anterior superior temporal lobe CSF volumes showed greater improvement. Large sulci in the posterior frontal and lateral temporal lobes reportedly predicted poorer response in another study.²²⁶ While clozapine decreased caudate volumes in patients previously treated with typical antipsychotics, this effect was reportedly unrelated to treatment response.²²⁷ It is reported that a larger right prefrontal gray matter volume in general was associated with better treatment response to clozapine but not to haloperidol.²²⁸ Patients with the greatest functional improvement reportedly had significantly less prefrontal sulcal widening.²²⁹ Larger volume in the dorsolateral prefrontal cortex at baseline may be associated with improvement in negative symptoms.²³⁰

Using SPECT, Molina Rodríguez et al.²³¹ found responders to clozapine at baseline had higher thalamic, left basal ganglia and right prefrontal perfusion while nonresponders had lower prefrontal perfusion, and responders showed higher subcortical perfusion than controls that decreased after clozapine treatment. Perfusion in the thalamus and right prefrontal regions was identified as response predictors by discriminant analysis.²³² Also a direct relation between clinical improvements and an increase in thalamic perfusion was observed.²³³ Using PET, Potkin et al.²³⁴ noted a relative increase of metabolic rates with clozapine in the basal ganglia especially on the right, and lowered metabolic rates in the frontal lobe more pronounced on the left as well as in the anterior nuclei of thalamus. It is also reported that clozapine, but not haloperidol, normalized blood flow patterns in the anterior cingulate during a cognitive task.²³⁵ Metabolic rates in the right inferior caudate at baseline reportedly differentiated clozapine and thiothixene responders.²³⁶ Clozapine resulted in recruitment of frontal, parietal and cingulate regions that did not appear to be active with high dose olanzapine in a case study.²³⁷ A series of PET perfusion studies indicated the following neural substrates for response: a high temporal gray matter volume for positive symptoms, a lower intracranial and hippocampal volume for disorganization symptoms, and a high baseline dorsolateral prefrontal cortical volume and metabolic activity for negative symptoms,²³⁰ a decrease of basal ganglia activity for negative symptoms, a metabolic decrease in motor area for disorganization, and an increase of activity in visual area for positive symptoms,²³⁸ and a clearly hypofrontal state after clozapine treatment.²³⁹

Combination Within/Across Factors

Those with 2664 C/C genotype in the ionotropic NMDA 2B subunit may require higher clozapine doses,^240,241 although it is also reported that genotype distribution was similar between patients and controls.²⁴¹ Carriers of 3435 T/T genotype in the ABCB1 that encodes P-glycoprotein transporter reportedly showed a 1.6-fold higher clozapine plasma concentration (Jaquenoud Sirot et al., 2009).⁸⁹ Consoli et al. (2009)²⁴² also reported that patients with 3435 C/C genotype required higher clozapine dose to achieve the same plasma concentration. A total of 6 out of 19 polymorphisms in serotonin and histamine receptors and serotonin transporters reportedly predicted 76.9% of responders.²⁴³ Lin et al.²⁴⁴ suggested an artificial neural network incorporating 5 genotypes including 5-HT2A, adrenergic and G-protein receptors together with 5 demographic factors, to predict response to clozapine with an overall accuracy of 83.3%.

EEG abnormalities with clozapine treatment are suggested to be blood-level dependent.^245,246 Negative correlations between clozapine serum levels and the amount of high-frequency EEG activity and positive correlations between high-frequency EEG activity and memory performance were reported.²⁴⁷ Gross et al.²⁴⁸ compared clozapine level and quantitative EEG to assess treatment adequacy and found that that midline electrodes over the fronto-central scalp area in quantitative EEG might be a more sensitive indicator for improvement. Yu et al.²⁴⁹ investigated ERPs and genotype in serotonin 5-HT2A receptor and found that those with 102C/C genotype had higher N100 amplitude than others after clozapine treatment. Nagamoto et al.²⁵⁰ reported an implication of improved P50 auditory gating as well as decrease in plasma MHPG levels for clozapine response.

Using SPECT, Ertugrul et al.²⁵¹ showed that clozapine increased bilateral frontal (superior and medial)/caudate perfusion ratios in responders. In addition, percentage changes in the left and right frontal (superior and medial)/caudate perfusion ratios compared to the baseline were higher in responders than nonresponders. They also revealed an increased N-acetyl-aspartate (NAA)/creatine plus phosphocreatine (Cre) ratio in the dorsolateral prefrontal cortex (DLPFC) after 8 weeks of clozapine treatment with ¹H-MRS. Potkin et al.²⁵² studied brain perfusion with PET as well as genotypes in serotonin and dopamine receptors in relation to clozapine response, and found that those with dopamine D1 receptor 2,2 genotype showed metabolic decreases in all major sectors of the brain with an exception of ventral parts of the caudate and putamen. Those with 1,2 genotype failed to show such changes and responded less to clozapine. However, a subsequent study found only a trend for the same direction on this genotype with respect to clozapine response.²⁵³

Discussion

This paper reviewed factors found to be associated with response to clozapine in schizophrenia. Despite a number of studies on this topic, no single factor appears to have shown adequate reproducibility, sensitivity and specificity for clozapine. Also, clinical feasibility is frequently problematic and cost-effectiveness of labor-extensive tests has not been addressed. Such an obvious difficulty may well be a reflection of the fact that pathophysiological bases of treatment resistance remain elusive. Clinical and demographic factors do not appear to be particularly useful in response prediction to clozapine, and higher symptoms scores before at baseline may simply indicate more room of improvement towards a regression to the mean (although they may indicate true resistance as well). On the other hand, it is suggested that an early response at 4 weeks may predict subsequent response²⁵ and it may not be yielding to wait beyond 8 weeks.²⁵⁴ However, another report indicated a majority of response occurred after 6 weeks and suggested a trial duration of 6–12 months.²⁵⁵ Further, persistence in response represents another important issue.²⁵⁶ Response can therefore be revisited with taking the timeline to treatment response into account.

Other factors are not conclusive as well. Although low HVA/5-HIAA ratio is proposed to serve as one of a very few predictors of response to clozapine,²⁵⁷ it has not been adequately addressed in recent studies. The current status of electrophysiological indices is they are not yet ready for clinical applications to improve diagnostic standards for schizophrenia²⁵⁸ and the same appears true with respect to response prediction. None of the genetic candidates has fully accounted for the etiology of schizophrenia and it appears unlikely that a contribution of a single genotype explains all of the prediction of response to any antipsychotics including clozapine.²⁵⁹ There are a number of negative findings as well on this topic^257,260 despite a possibility of publication bias regarding negative studies. It is indeed difficult to make a focus from ever expanding list of candidate genes²⁶¹ while ethnic stratification poses another challenge. From imaging perspectives, an adequate or even excessive blockade of dopamine D2 receptors, as evidenced by too high occupancy or a presence of EPS, may not be sufficient for clinical response to antipsychotics.^262,263 In fact, no evidence was found for a critical degree of D2 occupancy required to sustain therapeutic effects of clozapine,²⁶⁴ and patients did respond well to clozapine despite a lower D2 blockade.²⁶⁵ No support for regional selectivity with regards to D2 occupancy by clozapine is reported.²⁶⁶ Specificity of clozapine on 5-HT2A receptors is equivocal as well.²⁶⁷ Also still equivocal is a possible role of other receptors such as D1²⁶⁸ or D3.²⁶⁹ Nevertheless, the bases of treatment resistance have not been adequately explored beyond structural examinations and perfusion.

Finally, many studies reported on pharmacokinetic data. Still, considering a considerable variability in the response achieved at any given drug concentration, it would be prudent to titrate the dose upwards beyond 350–400 ng/ml in patients without side effects who failed to exhibit amelioration of psychopathology at standard dosages or at lower drug concentrations.⁷⁵ It is indeed reported that responders showed lower serum dose-corrected concentrations of clozapine compared to nonresponders²⁷⁰ and another study is also supportive for this notion,⁷⁸ although it is of course also possible that truly resistant patients remain unresponsive even with higher levels which poses another clinical challenge.²⁷¹ Variability is reportedly high with respect to norclozapine/clozapine ratio on both inter- and intra-individual bases.^272–274 Additionally, it is important to note that concentration-response relationships have not necessarily been positive.^67,275 It is reported that there was no difference in the incidence of adverse events in patients whose serum levels were below and those greater than the threshold of 200 ng/ml.⁷³ Moreover, the rate of side effects did not appear to be related to clozapine dose, clozapine or norclozapine plasma concentrations, or norclozapine/clozapine ratio in younger subjects, although the rate was generally higher than adult patients.⁷⁹ Further, it was found that blood level of clozapine did not predict central D2 blockade that remained consistently lower than typical antipsychotics, to conclude that careful clinical titration cannot be replaced by clozapine concentration monitoring.²⁷⁶ And most importantly, a significant variation in plasma levels did not necessarily result in a change in psychopathology²⁷⁷ and an increase in the level (by up to 58%) induced by fluoxetine had no impact on symptoms.²⁷⁸

Taken together, routine testing of serum clozapine levels has not been supported in everyday clinical practice, other than certain situations such as poor clinical response, signs of toxicity and suspected nonadherence.²⁷⁹ As such, we need treat the patient and not the level but a threshold of 350 ng/ml appears to be useful in case of nonresponse, although it is important to remember that response is well possible below this cutoff threshold.

It is likely that combination of the factors will gain more popularity in the future, although feasibility and cost-effectiveness remain to be an obvious obstacle.

Conclusion

Clozapine remains a drug of choice in treatment-resistant schizophrenia. However, in spite of a number of previous investigations, demographic and clinical variables have not been particularly useful in predicting response to clozapine. As for laboratory-based assessments, pharmacokinetic evidence implies a target clozapine level of 350 ng/ml but with substantial individual variability. Otherwise, limited evidence fails to extract clinically useful factors of response with certainty. It is important to note that in order for a factor to serve as a predictor of response, it is pre- and not post-clozapine results that are more pertinent,²⁵⁷ but the latter has been the main focus of many reports (i.e., investigation of after-results). Currently, not a single biological trait in schizophrenia is available which achieves sufficient specificity, selectivity and is based on causal pathology and predictive validity to be recommended as a diagnostic marker,²⁸⁰ a similar situation of which appears to be the case on factors and predictors for clozapine response. More studies are necessary on this important topic.