at SciVerse ScienceDirect

Neuropharmacology 64 (2013) 248e253

Contents lists available

Neuropharmacology

journal homepage: www.elsevier .com/locate/neuropharm

Predictors and mediators of add-on mirtazapine-induced cognitive enhancementin schizophrenia e A path model investigation

Jan-Henry Stenberg a,*, Viacheslav Terevnikov b, Marina Joffe a, Jari Tiihonen c,d,e, Evgeny Chukhin a,Mark Burkin f, Grigori Joffe a

aHelsinki University Central Hospital (HUCH), Helsinki, PO Box 750, FI-00029 HUCH, FinlandbKellokoski Hospital, Vanha Valtatie 198, FI-04500 Kellokoski, FinlandcDepartment of Clinical Neuroscience, Karolinska Institutet, S-171 77 Stockholm, SwedendUniversity of Eastern Finland, Niuvanniemi Hospital, FI-70240 Kuopio, FinlandeDepartment of Mental Health and Alcohol Research, National Institute for Health and Welfare, PO Box 30, FI-00271 Helsinki, FinlandfUniversity of Petrozavodsk, The University of Petrozavodsk, Department of Psychiatry, RU-185000 Petrozavodsk, Russia

a r t i c l e i n f o

Article history:Received 2 March 2012Received in revised form11 June 2012Accepted 14 June 2012

Keywords:Adjunctive mirtazapineFGARCTSchizophreniaCognitive enhancerPath analysis

* Corresponding author. Department of PsychiatryHospital (HUCH), PO Box 750, FI-00029 HUCH (Välssinki), Finland. Tel.: þ358 40 5198848 (mobile).

E-mail address: jan-henry.stenberg@hus.fi (J.-H. St

0028-3908/$ e see front matter � 2012 Elsevier Ltd.http://dx.doi.org/10.1016/j.neuropharm.2012.06.028

a b s t r a c t

We aimed to evaluate predictors and mediators of enhancing effect of adjunctive mirtazapine oncognition in schizophrenia. Patients with difficult-to-treat schizophrenia received either mirtazapine(n ¼ 19) or placebo (n ¼ 18) in a double-blind fashion for six weeks. Mirtazapine outperformed placeboon the Block Design and Stroop Dots. In the present subsidiary study, factors underlying this differencewere explored with Path Analysis. Add-on mirtazapine had an independent enhancing effect on theBlock Design-measured visuo-spatial functioning. Further, this effect was mediated via changes inpositive, depressive and parkinsonism symptoms, but not in negative symptoms. This effect was pre-dicted by higher doses of FGAs, longer duration of illness and lower initial Block Design scores. PathAnalysis model fit was good. Mirtazapine may have direct and indirect favorable effects on visuo-spatialfunctioning, but further research is needed. Path analysis may be a feasible statistical method for furtherresearch of neurocognition in psychopharmacological interventions in schizophrenia.

This article is part of a Special Issue entitled ‘Cognitive Enhancers’.� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Deficits in neurocognitive functioning in schizophrenia andtheir untoward effects on functional outcome have been welldocumented (Galletly, 2009), and enhancement of neurocognitivefunctions currently remains a principal goal in the treatment of thisdisease. As no a single antipsychotic agent has been consistentlyshown to be a neurocognitive enhancer, various adjunctivepsychopharmacological agents have been studied, yet so far withinconsistent and evasive results (Keefe et al., 2007a, 2007b;Galletly, 2009).

Some of these adjunctive agents are classified as antidepres-sants. Within this class of drugs, mirtazapine shows a unique, richmechanism of action that includes antagonism of noradrenaline a2,

, Helsinki University Centralkärinkatu 12, FI-00250 Hel-

enberg).

All rights reserved.

serotonin 5-HT2 and 5-HT3 receptors, and indirect agonism of the5-HT1A receptor (Anttila and Leinonen, 2001). All of these receptorsseem to be involved in the modulation of neurocognition(Akhondzadeh et al., 2009; Galletly, 2009; Sumiyoshi et al., 2007). Aproneurocognitive effect of adjunctive mirtazapine was first re-ported in an open label study by Delle Chiae et al. (2007), which hassubsequently been replicated in smaller randomized controlledtrials (RCT) (Stenberg et al., 2010, 2011; Cho et al., 2011).

In our previous studies, add-on mirtazapine not only enhancedthe antipsychotic effect of FGAs, and improved negative, mood andoverall symptomatology in chronic schizophrenia (Joffe et al., 2009;Terevnikov et al., 2010, 2011), but also had some desirable effects onneurocognition (Stenberg et al., 2010, 2011). Specifically, of the 21neurocognitive parameters tested, 8 parameters improved withstatistical significance within the mirtazapine group, while only 1parameter improved within the placebo group after a 6 weektreatment under double-blind conditions. Improved neuro-cognition differed between mirtazapine and placebo groups withstatistical significance on two parameters, the Block Design of theWechsler Adult Intelligence Scale e revised version (WAIS-R;

J.-H. Stenberg et al. / Neuropharmacology 64 (2013) 248e253 249

Wechsler, 1981), and Stroop Dots time (Jensen and Rohwer, 1966;Lezak, 1995).

While knowledge of the neurocognitive enhancing effects ofpsychopharmacological agents such as mirtazapine remainsincomplete, even less is known about the possible links betweenimpaired neurocognitive functions and the demographic or clinicalfeatures of patients with schizophrenia. Several studies (Heatonet al., 2001) [though not all (Klingberg et al., 2008)] suggest thatvariables such as high age, duration of illness and pronounceddyskinesia might be associated with impaired neurocognitivefunctioning (Heaton et al., 2001). Similar contradictions in thecurrent data exist in understanding of relationships between neu-rocognition and positive, negative, disorganized or affectivedomains of psychopathology of schizophrenia, as well as betweenlongitudinal changes in both clinical and neurocognitive variablesduring pharmacological treatments. To our knowledge, no studieshave focused on either predictors or mediators of the putativeproneurocognitive effects of mirtazapine in patients with schizo-phrenia. The current subsidiary study aimed to evaluate, whatfactors predict and what ones mediate the abovementionedmirtazapine-induced improvement in neurocognition. Thesecondary aim of the study was to evaluate the feasibility of pathanalysis for dissecting predictors from mediators in the complexconstellation of differing symptom domains, and analyzing theirinter-relationships over longitudinal effects of drug intervention.

2. Methods

This study was a part of an RCT where mirtazapine 30 mg/day or placebo wereadded to on-going FGAs in stable doses. The participants (n ¼ 19 in the mirtazapinegroup and n¼ 18 in the placebo group)were adults whowere suffering from chronicschizophrenia and had previously demonstrated sub-optimal response to differentFGAs. The study has been described thoroughly earlier (Joffe et al., 2009; Stenberget al., 2010).

The study was performed in the Matrosy Psychiatric Hospital and the DayTreatment Unit of the Psychoneurological Dispensary in Petrozavodsk, Russia. Thestudy protocol and its amendments were approved by the Ethics Committee of theKarelian Republic, Russian Federation and conducted in compliance with theDeclaration of Helsinki Guideline for Good Clinical Practice (ICH GCP), as well asnational and international regulations.

2.1. Assessments

Twenty one neurocognitive parameters were assessed with well establishedneuropsychological tests for a wide coverage of principal neurocognitive domains(Szöke et al., 2008).

In the present study, both predictors and mediators were explored for the twoneuropsychological tests that demonstrated statistically significant between groupdifferences; Block Design and Stroop Dots. The Block Design is a measure of visuo-spatial ability and fluency, while the Stroop Dots measures general mental speedand attention control. Age, sex, duration of illness, dose of antipsychotic medicationin chlorpromazine equivalents and baseline levels of the Block Design and StroopDots were evaluated as possible predictors of the improvement in neurocognition.

Variables that improved in the mirtazapine group in our clinical RCT (Joffe et al.,2009; Terevnikov et al., 2011) e the Positive and Negative Syndrome Scale (PANSS;Kay et al., 1987) and its subscales and the Calgary Depression Scale for Schizophrenia(CDSS; Addington et al., 1993) - were selected as possible mediators for themirtazapine-induced neurocognitive improvement. In addition, theSimpsoneAngus Scale (SAS) for Extrapyramidal Side Effects (Simpson and Angus,1970) was explored as possible mediator, since motoric symptoms are a part ofcognition.

2.2. Statistical analyses

The path analysis was employed for the identification of predictors and medi-ators. Path analysis allows a testing of a system of regression equations in whichvariables can serve as predictors, mediators and outcome. Thus, multiple outcomescan be studied in the same analysis. The procedure of pathway analysis allowseffects to be parsed apart in to their component (indirect, direct and total compo-nents) which is a significant advantage over multiple regressions. The estimatorused in the analyses is the robust maximum likelihood (MLR) estimator, whichcorrects for any non-normality in the variable distributions, and produces unbiased

estimates of the standard errors (Muthén andMuthén,1998e2010). Eachmodel wasfirst estimated by including all possible paths:

otherbaselinevariables

(predictors)

mirtazapine

mediator outcome

As the sample size was relatively small, the final model for each mediator-/outcome-combination was calculated with a minimal number of paths in themodel. This minimal number was reached by removing all paths from “otherbaseline variables” to the mediator and the outcome that were not statisticallysignificant in the initial model. For example, in most models, age and sex are notincluded in the final model at all, but they had been deleted from the modelsimply because their effect was not significant to begin with. With this samplesize, it was not sensable to test models that had more than 10 parameters, and allfinal models roughly had 10 parameters each. Due to small sample size, the resultswere both conservative for rejecting the null hypotheses and explorative. Forbalancing this effect in this small sample, a statistical significance level of p < 0.10was used, instead of the more frequently used p < 0.05. This definition wasdetermined a priori. The fact that the mirtazapine variable remained uncorrelated,relative to the other baseline variables, provides a statistically more efficientestimate of the paths emanating from this variable. Path analysis was performedusing Mplus 5.21.

Pearson’s Correlation Coefficient was employed to screen for possible effects ofmirtazapine-induced weight gain on cognition. This test was performed using IBMSPSS for Windows 19.0 software.

3. Results

The results for each explored mediator are given separately(Fig. 1). The paths mirtazapine/mediator, mirtazapine/outcome andmediator/outcome were always included in the final models,whether they were significant or not, as these paths are always oftheoretical interest. A solid line indicates that the path is statisticallysignificant, while a dashed line in the final model indicates that thepath was not statistically significant.

3.1. Block Design (Table 1)

3.1.1. Model for the PANSS negative subscaleMirtazapine treatment had an independent effect on both the

Block Design and negative symptoms, while the change in negativesymptoms did not mediate the improvement in Block Design. Abetter baseline level of Block Design functioning (i.e., higher points)predicted a smaller level of improvement (Table 1).

3.1.2. Model for the PANSS positive subscaleMirtazapine treatment had a direct effect on the Block Design

and PANSS positive scores. In addition, changes in PANSS posi-tive scores mediated the effect of mirtazapine on changes in theBlock Design. Higher values for changes in PANSS positive scoreswere associated with higher values in the Block Design. Forthis study population, this means that less improvement inPANSS positive scores meant more improvement in the BlockDesign.

Higher antipsychotic dose and shorter duration of illness wereassociated with larger improvement in Block Design.

3.1.3. Model for the PANSS total scaleMirtazapine treatment had an independent effect on both the

Block Design and PANSS total scores, but the change in the PANSStotal scores did not mediate the improvement seen in the BlockDesign.

Fig. 1. Path analysis of predictors and mediators of add-on mirtazapine-induced Block Design-measured cognitive enhancement in FGA-treated patients with schizophrenia.

J.-H. Stenberg et al. / Neuropharmacology 64 (2013) 248e253250

Better baseline functioning in the Block Design and longerduration of illness predicted a smaller improvement, while higherantipsychotic dosages predicted a larger improvement.

3.1.4. Model for the CDSSMirtazapine treatment was associated with an improvement in

the Block Design, both directly and via the CDSS. In other words,less improvement in CDSS scores meant more improvement in theBlock Design.

Better baseline functioning in the Block Design and longerduration of illness predicted smaller improvements in the BlockDesign, while higher antipsychotic dosages predicted a largerimprovement.

3.1.5. Model for the SASMirtazapine treatment was associated with an improvement in

the Block Design, both directly and via the SAS. In other words, lessimprovement in SAS scores meant more improvement in the BlockDesign.

Better baseline function in the Block Design and longer durationof illness predicted a smaller improvement in the Block Design,whilehigher antipsychotic dosages predicted a larger improvement.

3.2. Stroop Dots (data not shown)

Mirtazapine treatment was not independently associated withan improvement in the Stroop Dots for any model.

In all models (except the one for the SAS), higher baselineperformance (i.e., lower time on Stroop Dots) and higher dosages ofantipsychotics predicted a smaller improvement (i.e., a smallerdrop in time) on the Stroop Dots. Of all parameters measured, onlythe SAS mediated the effects of mirtazapine on the improvement inthe Stroop Dots. In other words, decreases in the SAS scores werecorrelated with decrease in time in the Stroop Dots.

3.3. Model fit

Overall model fit of the explored path models was good, i.e., thestatistical differences between the models and data were notsignificant in any model.

Mirtazapine but not placebo inducedweight gain (mean 3.24 kg,p ¼ 0.000 vs. 0.04 kg, p ¼ 0.918, correspondingly). Weight gaincorrelated with neither changes in Block Design (r ¼ 0.056,p ¼ 0.756) nor changes in Stroop Dots (r ¼ �0.127, p ¼ 0.466).

Table 1Predictors and mediators of add-on mirtazapine-induced enhancement in BlockDesign measured cognitive function in schizophrenia: results of path analysis.

Path analysismodel

Predictor ormediator

Value, D (s.e.) t ¼ p ¼(two-tailed)

Overallmodel fit,p ¼

Model for PANSS negative scores 0.845Block Design, D mirtazapine 2.87 (1.58) 1.81 0.070

Block Design,baseline

�0.18 (0.08) �2.39 0.017

PANSSn, D �0.29 (0.29) �0.98 0.328Model for PANSS positive scores 0.791Block Design, D mirtazapine 3.73 (1.09) 3.44 0.001

duration ofillness

�0.23 (0.06) �3.68 0.000

FGA dose 0.01 (0.01) 2.29 0.022Block Design,baseline

�0.22 (0.08) �2.85 0.004

PANSSp, D 0.35 (0.13) 2.82 0.005Model for PANSS total scores 0.492Block Design, D mirtazapine 2.99 (1.06) 2.82 0.005

duration ofillness

�0.20 (0.06) �3.17 0.002

FGA dose 0.01 (0.01) 2.34 0.020Block Design,baseline

�0.21 (0.08) �2.59 0.010

PANSSt, D 0.04 (0.04) 1.00 0.317PANSSt, D mirtazapine �10.30 (3.18) �3.24 0.001Model for CDSS 0.490Block Design, D mirtazapine 3.17 (1.09) 2.92 0.004

duration ofillness

�0.20 (0.06) �3.38 0.001

FGA dose 0.01 (0.01) 1.92 0.054Block Design,baseline

�0.19 (0.07) �2.62 0.009

CDSS, D 0.30 (0.12) 2.39 0.017CDSS, D mirtazapine �1.85 (0.95) �1.94 0.051Model for SAS 0.518Block Design, D mirtazapine 3.50 (1.09) 3.22 0.001

duration ofillness

�0.15 (0.06) �2.77 0.006

FGE dose 0.01 (0.00) 2.08 0.038Block Design,basline

�0.20 (0.07) �2.89 0.004

SAS, D 0.41 (0.20) 2.08 0.037SAS, D mirtazapine �2.04 (0.97) �2.10 0.036

� D ¼ change (scores at week 0 minus scores at week 6).� PANSS ¼ Positive and Negative Syndrome Scale.� PANSSn ¼ scores for negative subscale of the PANSS.� PANSSp ¼ scores for positive subscale of the PANSS.� PANSSt ¼ PANSS totals scores.� CDSS ¼ scores for Calgary Depression Scale for Schizophrenia.� SAS ¼ scores for Simpson-Angus Scale for Extrapyramidal Side Effects.� FGA ¼ First Generation Antipsychotics (dose in chlorpromazine equivalents).� For balancing the effect of small sample size (i.e., results are both conservative forrejecting null hypotheses and explorative), the significance level of p < 0.10 insteadof frequently used p < 0.05 was a priori defined. Here, statistically significantfindings are expressed in bold.� Overall model fit ¼ chi-square, significant means statistical differences betweenthe models and data (i.e., weak fit) and non-significant means no statisticallysignificant differences between the models and data (i.e., good fit).

J.-H. Stenberg et al. / Neuropharmacology 64 (2013) 248e253 251

4. Discussion

In the present study, mirtazapine treatment demonstrateda noticeable (35e38%) independent direct favorable effect on thevisuo-spatial ability as measured with the Block Design in all pathanalysis models. Moreover, changes in the PANSS positive, CDSSand SAS scores mediated the effect of mirtazapine on the BlockDesign, while changes in the PANSS negative or PANSS total scoresdid not. Higher dosages of antipsychotics and shorter duration ofillness predicted more noticeable improvement in the Block Designin all models except the one for PANSS negative scores. A betterbaseline level for the Block Design predicted less improvement

with adjunctive mirtazapine treatment. Further, mirtazapine didnot have an independent effect on the Stroop Dots, while higherdosages of FGAs did predict and change in the SAS scores didmediate the effect of mirtazapine treatment on the Stroop Dots.

4.1. Mediators of mirtazapine-induced improvement in the BlockDesign

Interestingly, while the PANSS positive, CDSS and SAS scores didimprove in these patients, the relationship of these scores withimprovement in the Block Design was rather reverse. In otherwords, less improvement in these clinical measures meant moreimprovement in the Block Design.

The interrelationship between neurocognitive deficits and othersymptoms of schizophrenia is not entirely understood. Especiallynegative symptoms have been the focus of neuropsychologicaldiscussions, as they tend to overlap with neurocognitive deficits.Some researchers have assumed neurocognitive deficits to besecondary to clinical symptoms (Kerns and Berenbaum, 2002;Nieuwenstein et al., 2001), while others have maintained that theyare an independent domain of schizophrenia (Green, 1996; Harveyet al., 2006; Dominguez et al., 2009). A recent systematic review(Dominguez et al., 2009) and a later publication by Ventura et al.(2009) showed that negative symptoms were significantly associ-ated with cognitive deficits, although modestly in a statisticalsense, while positive and depressive dimensions of psychopa-thology were not. Nevertheless, in the Clinical Antipsychotic Trialsof Intervention Effectiveness (CATIE) study published by Keefe et al.(2007a), neurocognitive improvement was clearly, althoughmodestly, associated with symptom reduction in all PANSSdimensions.

In the present study, the absence of any significant relationshipbetween the improvement in neurocognition with that in negativesymptoms and, on the other hand, the mediation of a proneur-ocognitive effect of mirtazapine via changes in positive, depressive,and parkinsonism symptoms, might be due to the specific featuresof our study population. Our chronically ill and highly psychoticpatients showed the highest (or fastest) improvement in positiveand depressive symptoms (Joffe et al., 2009; Terevnikov et al., 2010,2011)e themagnitude of change that could just bemost easily seenin the statistical calculations. Moreover, the seemingly inverserelationship between improvement in the Block Design and that inthe positive, depressive and parkinsonism symptoms might be dueto purely statistical reasons, too. In other words, initially low clin-ical symptom scores in some patients could only allow for a minordrop in clinical ratings, as the result of a ceiling effect, while thecapacity of neurocognitive improvement in these patients could behigh or even higher than that of their more symptomatic coun-terparts. The possible role of clinical mediators in the effects ofmirtazapine on neurocognition thus remains at least partly unclear,and requires more studies that employ the statistical method ofpath analysis in both different and larger patient populations.

As expected, mirtazapine induced weight gain, but the drug’seffect on body weight did not correlate with changes in cognition.Weight gain was, thus, unlikely a mediator of the mirtazapine-induced cognitive enhancement.

4.2. Predictors of mirtazapine-induced improvement in the BlockDesign

In schizophrenia, the FGAs tend to impair neurocognition, atleast when high dosages are used (Keefe et al., 2007a). In thepresent study, higher dosages of FGAs predicted a desired effectfrommirtazapine on the Block Design. In other words, mirtazapine

J.-H. Stenberg et al. / Neuropharmacology 64 (2013) 248e253252

seemed to remediate the adverse neurocognitive effect of FGAs onneurocognition.

According to a recent meta-analysis (Szöke et al., 2008), theremaining capacity of neurocognitive rehabilitation in patientswith prolonged schizophrenia is limited. Therefore, it was probablynot surprising to find in this study that the duration of illnessnegatively correlated with mirtazapine-induced neurocognitiveimprovement in most path analysis models.

Those patients who had a better baseline level for the BlockDesign showed less improvement with adjunctive mirtazapinetreatment, probably due to a ceiling effect. Conversely, this mightmean that patients with a more prominent schizophrenia-inducedimpairment in visual-spatial functioning could benefit even morefrom add-on mirtazapine treatment.

4.3. Stroop Dots

Mirtazapine seemed to enhance general mental speed andattention control, probably indirectly, due to improvement in theFGA-related symptoms of parkinsonism.

4.4. Limitations and future research

The limitations of this study were primarily the small samplesize and an inability to use a composite score for neurocognitivetests, thereby limiting the analyses to only two tests (Stenberg et al.,2010, 2011) and a specific patient population. Thus, one shouldremain cautious when extrapolating our results to areas other thanvisuo-spatial domains of neurocognition, FGAs antipsychotics, orpatient populations other than chronic and highly symptomaticschizophrenic patients with schizophrenia. To the best of theauthors’ knowledge, this is the first report on path analysis as anobjective method for the exploration of possible relationshipsbetween psychopharmacological interventions and related changesin neurocognition in schizophrenia.

The statistical differences between the models and data werenot significant in any model, meaning that the model fit of theexplored path models was good. Thus, path analysis appears to bea feasible method for an objective investigation of predictors andmediators of potential cognitive enhancers in schizophrenicpatients, which was a secondary aim of this study.

The question of primacy of cognitive deficits in the complexconstellation of psychosis, negative symptoms, cognitive impair-ment and other psychopathology is an issue of debate (Galletly,2009). Since this research focused on cognition, cognitive changeswere chosen as outcome and other variables of interest as potentialpredictors or mediators. It is possible, however, that the drug’seffects on symptoms were secondary to its effects on cognition.Moreover, the drug could improve contemporaneously clinicalsymptoms and cognition indirectly and independently, by reac-tivating neuroplasticity (Fumagalli et al., 2009) or via anotherunknown third factor.

5. Conclusions

The favorable effect of mirtazapine added to FGAs on visuo-spatial functioning in schizophrenia may be direct; i.e., inde-pendent of the mirtazapine-induced improvement in otherclinical domains of schizophrenia. In addition, this beneficialeffect of mirtazapine on visuo-spatial functioning, as measuredwith Block Design, may be mediated via changes in positive,depressive and parkinsonism symptoms, but not in negativesymptoms, and may be predicted by higher dosages of FGAs,longer duration of illness and lower initial Block Design scores.Path analysis may be a feasible statistical method for further

research on neurocognition in psychopharmacological inter-ventions in schizophrenia.

Acknowledgments

This trial (01T-67) was supported by a grant from the StanleyMedical Research Institute (SMRI), Bethesda, MD, USA. The SMRIhad no role in either the study design, the gathering or analysis ofdata, or in the writing or publication of this report. The study wasregistered by Current Controlled Trials (http://www.controlled-trials.com, trial registration number ISRCTN00721331).

References

Addington, D., Addington, J., Maticka-Tindale, E., 1993. Assessing depression inschizophrenia: the Calgary depression scale. Br. J. Psychiatry 163, 39e44.

Akhondzadeh, S., Mohammadi, N., Noroozian, M., Karamghadiri, N., Ghoreishi, A.,Jamshidi, A.H., Forghani, S., 2009. Added ondansetron for stable schizophrenia:a double blind, placebo controlled trial. Schizophr. Res. 107, 206e212.

Anttila, S.A., Leinonen, E.V., 2001. A review of the pharmacological and clinicalprofile of mirtazapine. CNS Drug Rev. 7, 249e264.

Delle Chiae, R., Salviati, M., Fiorentini, S., Biondi, M., 2007. Add-on mirtazapineenhances effects on cognition in schizophrenic patients under stabilizedtreatment with clozapine. Exp. Clin. Psychopharmacol. 15, 563e568.

Cho, S.J., Yook, K., Kim, B., Choi, T.K., Lee, K.S., Kim, Y.W., Lee, J.E., Suh, S., Yook, K.H.,Lee, S.H., 2011. Mirtazapine augmentation enhances cognitive and reducesnegative symptoms in schizophrenia patients treated with risperidone:a randomized controlled trial. Prog. Neuropsychopharmacol. Biol. Psychiatry 35,208e211.

Dominguez, M., Viecchtbauer, W., Simons, C.J.P., van Os, J., Krabbendam, L., 2009.Are psychotic psychopathology and neurocognition orthogonal? A systematicreview of their associations. Psychol. Bull. 135, 157e171.

Fumagalli, F., Frasca, A., Racagni, G., Riva, M.A., 2009. Cognitive effects of second-generation antipsychotics: current insights into neurochemical mechanisms.CNS Drugs 23, 603e614.

Galletly, C., 2009. Recent advances in treating cognitive impairment in schizo-phrenia. Psychopharmacology 202, 259e273.

Green, M.F., 1996. What are the functional consequences of neurocognitive deficitsin schizophrenia? Am. J. Psychiatry 153, 321e330.

Harvey, P.D., Koren, D., Reichenberg, A., Bowie, C.R., 2006. Negative symptoms andcognitive deficits: what is the nature of their relationship? Schizophr. Bull. 32,250e258.

Heaton, R.K., Gladsjo, J.A., Palmer, B.W., Kuck, J., Marcotte, T.D., Jeste, D.V., 2001.Stability and course of neuropsychological deficits in schizophrenia. Arch. Gen.Psychiatry 58, 24e32.

Jensen, A.R., Rohwer Jr., W.D., 1966. The Stroop Color-Word test: a review. ActaPsychol. (Amst.) 25, 36e93.

Joffe, G., Terevnikov, V., Joffe, M., Stenberg, J.H., Burkin, M., Tiihonen, J., 2009. Add-on mirtazapine enhances antipsychotic effect of first generation antipsychoticsin schizophrenia: a double-blind, randomized, placebo-controlled trial. Schiz-ophr. Res. 108, 245e251.

Kay, S., Fiszbein, A., Opler, L., 1987. The positive and negative syndrome scale(PANSS) for schizophrenia. Schizophr. Bull. 13, 261e276.

Keefe, R.S.E., Bilder, R.M., Davis, S.M., Harvey, P.D., Palmer, B.W., Gold, J.M.,Meltzer, H.Y., Green, M.F., Capuano, G., Stroup, T.S., McEvoy, J.P., Swartz, M.S.,Rosenheck, R.A., Perkins, D.O., Davis, C.E., Hsiao, J.K., Lieberman, J.A., CATIEInvestigators; Neurocognitive Working Group, 2007a. Neurocognitive effects ofantipsychotic medications inpatients with chronic schizophrenia in the CATIEtrial. Arch. Gen. Psychiatry 64, 633e647.

Keefe, R.S.E., Sweeney, J.A., Gu, H., Hamer, R.M., Perkins, D.O., McEvoy, J.P.,Lieberman, J.A., 2007b. Effects of olanzapine, quetiapine, and risperidone onneurocognitive function in early psychosis: a randomized, double-blind 52-week comparison. Am. J. Psychiatry 164, 1061e1071.

Kerns, J.G., Berenbaum, H., 2002. Cognitive impairments associated with formalthought disorder in people with schizophrenia. J. Abnorm. Psychol. 111,211e224.

Klingberg, S., Wittorf, A., Sickinger, S., Buchkremer, G., Wiedemann, G., 2008. Courseof cognitive functioning during the stabilization phase of schizophrenia.J. Psychiatr. Res. 42, 259e267.

Lezak, M.D., 1995. Neuropsychological Assessment, third ed. Oxford UniversityPress, New York.

Muthén, L.K., Muthén, B.O., 1998e2010. Mplus User’s Guide, sixth ed. Muthén &Muthén, Los Angeles, CA.

Nieuwenstein, M.R., Aleman, A., de Haan, E.H., 2001. Relationship betweensymptom dimensions and neurocognitive functioning in schizophrenia:a meta-analysis of WCST and CPT studies. J. Psychiatr. Res. 35, 119e125.

Simpson, G., Angus, J., 1970. A rating scale for extrapyramidal side effects. ActaPsychiatr. Scand. 45 (Suppl. 212), 11e12.

Sumiyoshi, T., Park, S., Jayathilake, K., Roy, A., Ertugrul, A., Meltzer, H.Y., 2007. Effectof buspirone, a serotonin 1A partial agonist, on cognitive functioning in

J.-H. Stenberg et al. / Neuropharmacology 64 (2013) 248e253 253

schizophrenia: a randomized double-blind placebo-controlled study. Schizophr.Res. 95, 158e168.

Stenberg, J.-H., Terevnikov, V., Joffe, M., Tiihonen, M., Tchoukhine, E., Burkin, M.,Joffe, G., 2010. Effects of add-on mirtazapine on neurocognition in schizo-phrenia: a double-blind, randomized, placebo-controlled study. Int. J. Neuro-psychopharmacol. 13, 433e441.

Stenberg, J.-H., Terevnikov, V., Joffe, M., Tiihonen, J., Tchoukhine, E., Burkin, M.,Joffe, G., 2011. More evidence on proneurocognitive effects of add-on mirtaza-pine in schizophrenia. Prog. Neuropsychopharmacol. Biol. Psychiatry 35,1080e1086.

Szöke, A., Trandafir, A., Dupont, M.E., Méary, A., Schürhoff, F., Leboyer, M., 2008.Longitudinal studies of cognition in schizophrenia :meta-analysis. Br. J.Psychiatry 192, 248e257.

Terevnikov, V., Stenberg, J.-H., Joffe, M., Tiihonen, J., Burkin, M., Tchoukhine, E.,Joffe, G., 2010. More evidence on additive antipsychotic effect of adjunctivemirtazapine in schizophrenia: an extension phase of a randomized controlledtrial. Hum. Psychopharmacol. Clin. Exp. 25, 431e438.

Terevnikov, V., Stenberg, J.-H., Tiihonen, J., Joffe, M., Burkin, M., Tchoukhine, E.,Joffe, G., 2011. Add-on mirtazapine improves depressive symptoms in schizo-phrenia: a double-blind randomized placebo-controlled study with an open-label extension phase. Hum. Psychopharmacol. Clin. Exp. 26, 188e193.

Ventura, J., Hellemann, G.S., Thames, A.D., Koellner, V., Nuechterlein, K.H., 2009.Symptoms as mediators of the relationship between neurocognition and func-tional outcome in schizophrenia: a meta-analysis. Schizophr. Res. 113, 189e199.

Wechsler, D., 1981. Wechsler Adult Intelligence Scale e Revised. The PsychologicalCorporation, San Antonio, TX.