Improvement in mismatch negativity generation during d -serine treatment in schizophrenia: Correlation with symptoms

SCHRES-07190; No of Pages 10 Schizophrenia Research xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Schizophrenia Research journal homepage: www.elsevier.com/locate/schres

Improvement in mismatch negativity generation during D-serine treatment in schizophrenia: Correlation with symptoms Joshua T. Kantrowitz a,b,⁎, Michael L. Epstein a,b,d, Migyung Lee a,b, Nayla Lehrfeld a, Karen A Nolan a,c, Constance Shope a, Eva Petkova a,e, Gail Silipo a, Daniel C. Javitt a,b a

Schizophrenia Research Center, Nathan Kline Institute, 140 Old Orangeburg Road, Orangeburg, NY 10962, United States Department of Psychiatry, Columbia University, 1051 Riverside Drive, New York, NY 10032, United States Department of Psychiatry, New York University School of Medicine, 1 Park Ave, New York, NY, United States d Graduate Center, City University of New York, 365 5th Ave, New York, NY, United States e Department of Child and Adolescent Psychiatry, New York University School of Medicine, 1 Park Ave, New York, NY, United States b c

a r t i c l e

i n f o

Article history: Received 30 January 2017 Received in revised form 24 February 2017 Accepted 27 February 2017 Available online xxxx

a b s t r a c t Background: Deficits in N-methyl-D-aspartate-type (NMDAR) function contribute to symptoms and cognitive dysfunction in schizophrenia. The efficacy of NMDAR agonists in the treatment of persistent symptoms of schizophrenia has been variable, potentially reflecting limitations in functional target engagement. We recently demonstrated significant improvement in auditory mismatch negativity (MMN) with once-weekly treatment with D-serine,

a naturally occurring NMDAR glycine-site agonist. This study investigates effects of continuous (daily) NMDAR agonists in schizophrenia/schizoaffective disorder. Methods: Primary analysis was on MMN after double-blind crossover (60 mg/kg/d, n = 16, 6 weeks) treatment with D-serine/placebo. Secondary measures included clinical symptoms, neurocognition, and the effects of openlabel (30–120 mg/kg/d, n = 21) D-serine and bitopertin/placebo (10 mg, n = 29), a glycine transport inhibitor. Results: Double-blind D-serine treatment led to significant improvement in MMN frequency (p = 0.001, d = 2.3) generation and clinical symptoms (p = 0.023, d = 0.80). MMN frequency correlated significantly with change in symptoms (r = −0.63, p = 0.002) following co-variation for treatment type. D-Serine treatment led to a significant, large effect size increase vs. placebo in evoked α-power in response to standards (p = 0.036, d = 0.81), appearing to normalize evoked α power relative to previous findings with controls. While similar results were seen with open-label D-serine, no significant effects of bitopertin were observed for symptoms or MMN. Conclusions: These findings represent the first randomized double-blind placebo-controlled study with 60 mg/kg D-serine in schizophrenia, and are consistent with meta-analyses showing significant effects of D-serine in schizophrenia. Results overall support suggest that MMN may have negative, as well as positive, predictive value in predicting efficacy of novel compounds. Clinical trials registration: Clinicaltrials.gov: NCT00322023/NCT00817336 (D-serine); NCT01116830 (bitopertin). © 2017 Elsevier B.V. All rights reserved.

1. Introduction Currently approved treatments for schizophrenia, including both typical and atypical antipsychotics, function primarily as dopamine (D2) receptor antagonists. While effective in treatment of positive symptoms, antipsychotics have limited efficacy against persistent negative symptoms or cognitive impairments, necessitating the development of alternative treatment approaches (Insel, 2010; Javitt, 2015a). Recent neurochemical models of persistent dysfunction in schizophrenia focus on disturbances of brain glutamatergic neurotransmission, based upon the ability of ⁎ Corresponding author at: Nathan Kline Institute, 140 Old Orangeburg Road, Orangeburg, NY 10962, United States. E-mail address: [email protected] (J.T. Kantrowitz).

phencyclidine (PCP), ketamine and similar compounds to induce deficits closely resembling those of schizophrenia by blocking neurotransmission at N-methyl-D-aspartate-type glutamate receptors (NMDAR) (Coyle, 2012; Javitt and Zukin, 1991; Kantrowitz and Javitt, 2010a, 2010b; Krystal et al., 1994; Moghaddam and Krystal, 2012). Despite the wellreplicated ability of NMDAR antagonists to induce schizophrenia-like symptoms and neurophysiological deficits (Krystal et al., 1994; Umbricht et al., 2000), the ability of NMDAR agonists to reverse such deficits remains unknown (Javitt et al., 2011). Moreover, the efficacy of NMDAR agonists in the treatment of persistent symptoms of schizophrenia has been variable across studies to date (Buchanan et al., 2007; Tsai and Lin, 2010; Weiser et al., 2012), potentially reflecting limitations in functional target engagement. The principal goal of this project was to assess the ability of neurophysiological measures such as auditory mismatch

Please cite this article as: Kantrowitz, J.T., et al., Improvement in mismatch negativity generation during d-serine treatment in schizophrenia: Correlation with symptoms, Schizophr. Res. (2017), http://dx.doi.org/10.1016/j.schres.2017.02.027

2

J.T. Kantrowitz et al. / Schizophrenia Research xxx (2017) xxx–xxx

(MMN) (Naatanen et al., 2014) to assess the functional target engagement of NMDAR modulating drugs for translational drug development (Javitt et al., 2011; Javitt and Sweet, 2015). 1.1. MMN as a biomarker for NMDAR dysfunction in schizophrenia As opposed to dopaminergic models, glutamatergic models specifically account for the impaired generation of MMN, a neurophysiological biomarker elicited most commonly in the context of an auditory oddball paradigm, in which a sequence of repetitive standards is interrupted infrequently by a physically different oddball stimulus. Deviants differ from standards in one or more physical and/or abstract dimensions, including frequency, intensity, duration or spatial localization (Javitt, 2000; Mantysalo and Naatanen, 1987). MMN is increasingly conceptualized as reflecting the “prediction error” engendered by the deviant vs. standard stimulus, which, in turn, is tied to NMDAR function at the level of auditory sensory cortex (Friston, 2005; Garrido et al., 2009; Todd et al., 2012; Wacongne, 2016; Wacongne et al., 2012). MMN generation has been shown to be sensitive to NMDAR dysfunction in both monkey models (Javitt et al., 1996) and in healthy human volunteers (Gunduz-Bruce et al., 2012; Javitt et al., 2011; Rosburg and Kreitschmann-Andermahr, 2016; Rowland et al., 2016; Umbricht et al., 2000). By contrast, MMN dysfunction is neither induced by dopamine agonists (Leung et al., 2007) or related psychotomimetic agents [e.g. psilocybin, LSD (Heekeren et al., 2008)] nor reversed by dopamine antagonists such as risperidone or clozapine (Umbricht et al., 1998b) and other typical or atypical antipsychotics (Umbricht et al., 1998a, 1999), suggesting relative specificity for glutamatergic vs. dopaminergic mechanisms. Deficits in auditory MMN generation in schizophrenia were first demonstrated in the early 1990′s and have been replicated extensively since that time (Erickson et al., 2015; Friedman et al., 2012; Green et al., 2012; Javitt et al., 2008b; Light et al., 2012; Light et al., 2014; Umbricht and Krljes, 2005), and tied to poor functional outcome in schizophrenia (Jahshan et al., 2013; Javitt and Freedman, 2015; Kantrowitz et al., 2015a; Light et al., 2014; Thomas et al., 2016), supporting the relevance of sensory-level dysfunction to new treatment development. In addition, test-retest reliability of MMN is high, encouraging its use as a neurophysiological biomarker for new treatment development (Javitt et al., 1996; Light and Swerdlow, 2015; Light et al., 2012). 1.2. Rationale for study of D-serine The present study primarily investigates MMN prior to and following double-blind treatment with D-serine, administered at a dose of 60 mg/kg (~4 g/d) for 6 weeks, along with a secondary analysis of the biomarker results of our open-label, dose escalation study (Kantrowitz et al., 2010) and bitopertin (Section 1.4). D-Serine functions as an endogenous ligand for the glycine modulatory site of the NMDAR (Balu and Coyle, 2015). A recent meta-analysis suggests that basal D-serine levels are reduced in schizophrenia (Cho et al., 2016), potentially related to genetic variation of genes such as serine racemase and D-amino acid oxidase (DAAO), which are involved in the synthesis and degradation of D-serine, respectively. The majority of studies with D-serine have used a dose of 30 mg/kg (~2 g/d), with significant, but small (SMD = −0.3) beneficial effects on positive and negative schizophrenia symptoms across studies (Cho et al., 2016). We have demonstrated safety of a higher dose (60 mg/kg: ~4 g/d) of D-serine in an open-label, dose escalation study (Kantrowitz et al., 2010), as well as significant, moderate-size (d = 0.68) beneficial effects in a study of 44 individuals at clinically high risk for schizophrenia (Kantrowitz et al., 2015b). We have also recently demonstrated that intermittent (once-weekly) treatment with D-serine over 2–3 weeks combined with auditory plasticity training leads to significant acute improvement in both plasticity and MMN in schizophrenia (Kantrowitz et al., 2016), without significant side effects.

1.3. Event-related potential (ERP) and time-frequency analysis of MMN For the present study, MMN was obtained using an “optimized” paradigm with intermixed frequency, duration and intensity deviants, as described previously (Friedman et al., 2012). In the present project, MMN was analyzed using time-domain event-related potential (ERP) (Javitt, 2015b; Javitt et al., 2008a; Luck et al., 2011) and using an exploratory analysis of time-frequency decomposition. In ERP, MMN is manifested as a negative response over frontocentral scalp, while in timefrequency, event-related activity is divided into discrete θ (4–7 Hz), α (7–12 Hz), β (12–24 Hz) and γ (N24 Hz) bands. Within these bands, stimulus-related activity is further differentiated into those that reflect alterations in phase reset mechanisms—intertrial coherence (ITC) vs. those that reflect alterations in single-trial power (e.g. Lakatos et al., 2013; Lisman et al., 2008). Most MMN analysis using ERP focuses on the deviant vs. standard differences, as the standard stimuli themselves are difficult to resolve using ERP. Recent human (Hsiao et al., 2009; Potes et al., 2014) and primate (Haegens et al., 2015) intracranial recording studies of thalamocortical activity suggest that response to standards occurs primarily within the α-frequency band. In addition, we (Javitt et al., 2000; Lakatos et al., 2013; Lee et al., 2017), and others (Hong et al., 2012; Kayser et al., 2014; Ko et al., 2012) have previously linked MMN dysfunction in schizophrenia primarily to impaired θ-frequency response, which may index function within corticocortical networks (Hsiao et al., 2009) involving somatostatin (SST)-type interneurons (Javitt and Sweet, 2015; Womelsdorf et al., 2014). The present report's analysis of the time-frequency signal of MMN is guided by these prior studies, and our recently published work (Kantrowitz et al., 2016; Lee et al., 2017). In our trial of intermittent Dserine treatment (Kantrowitz et al., 2016), we noted significant acute improvement in θ ITC during plasticity training, but this will be the 1st examination of D-serine effects on the time-frequency signal of MMN. Moreover, we recently demonstrated that schizophrenia patients have both robust deficits in α-frequency and elevated θ-frequency (reduced suppression) in response to standards (Lee et al., 2017). By contrast, during MMN, schizophrenia patients exhibited θ-frequency deficits and elevated α-frequency (reduced suppression). Moreover, responses in the θ-frequency response to both standards and MMN were predictive of symptoms. 1.4. Clinical outcomes, meta-analysis, open-label and bitopertin Along with biomarkers, we evaluate effects of D-serine on symptoms using the Positive and Negative Symptom Scale (PANSS) and on cognition using the MATRICS consensus cognition battery (MCCB). We hypothesized that agents would improve symptoms of schizophrenia to the extent that they induced objective benefit in neurophysiological biomarkers. In order to evaluate the present results in the context of prior studies (Cho et al., 2016), an updated meta-analysis of the effect of D-serine (along with the related compound D-alanine) on negative symptoms was conducted (Section 3.6). Finally, as a contrast to double-blind D-serine results, we present the results of two previously presented studies: open-label D-serine (Kantrowitz et al., 2010) and bitopertin (Da Costa et al., 2015; Kantrowitz et al., submitted for publication) conducted by the same investigators and using the same equipment and neurophysiological paradigm. Full methodological details on these trials are published separately. These trials used identical symptomatic, cognitive and biomarker methodology, allowing for easy comparison, and are included to assess whether MMN may have negative, as well as positive predictive value. Trial design and clinical/neurocognitive (but not biomarker) results of open-label D-serine study have been published previously (Kantrowitz et al., 2010). Bitopertin is a recently developed high affinity glycine type I (GlyT1) transport inhibitor (Alberati et al., 2012). In an initial phase II study bitopertin showed significant beneficial effects

Please cite this article as: Kantrowitz, J.T., et al., Improvement in mismatch negativity generation during d-serine treatment in schizophrenia: Correlation with symptoms, Schizophr. Res. (2017), http://dx.doi.org/10.1016/j.schres.2017.02.027

J.T. Kantrowitz et al. / Schizophrenia Research xxx (2017) xxx–xxx

(Umbricht et al., 2014), but follow-up phase III results were mixed (Arango et al., 2014; Blaettler et al., 2014; Bugarski-Kirola et al., 2016). 2. Materials and methods 2.1. Subjects/design (Supplemental Fig. 1A/B/C) All projects were conducted at the Nathan Kline Institute, Orangeburg, NY using identical methods and analysis, unless specifically noted. Presented results are from the double-blind cross-over D-serine study, other than open-label D-serine and bitopertin results in Section 3.8. Written informed consent for participation was obtained from all subjects. Subjects were aged 18–64 with DSM-IV SCID diagnosis of schizophrenia or schizoaffective disorder (First et al., 1997). Functional impairment was determined by a PANSS negative symptom subscale score N 20 for two consecutive weeks (Kay et al., 1987). Subjects were psychiatrically stable as evidenced by a Clinical Global impression (CGI) Improvement score (Guy, 1976) of 3 or 4 in the two weeks post screening visit. Subjects were excluded for unstable medical illness or any renal impairment [Glomerular Filtration Rate (GFR) ≥ 60; D-serine only]; alcohol/substance abuse within past month or dependence within past six months; extrapyramidal side effects (EPS) defined by Simpson Angus Scale (Simpson and Angus, 1970) total score ≥ 12; depression as defined by Calgary Depression Inventory for Schizophrenia (Addington et al., 1994) total score ≥ 10; or suicidal ideation. Patients on clozapine were excluded because of reports of reduced efficacy in combination with NMDA modulators (Tsai et al., 1999). Subjects had no psychotropic medication changes within the prior 4-weeks. 2.1.1. Double-blind D-serine Subjects received a fixed dose of 60 mg/kg/day in a randomized, placebo-controlled crossover study of D-serine and placebo each for 6 weeks. Subjects were randomly assigned to receive (D-serine or placebo) in the first treatment phase in a 1:1 ratio, followed by two-weeks of single-blind placebo washout, which was followed by the alternative treatment in a second-phase. In our previously published open-label study (Kantrowitz et al., 2010), a two-week washout was sufficient to reduce D-serine to baseline levels. Two double-blind subjects withdrew consent prior to post baseline outcome assessments and 3 active and 3 placebo subjects were removed from MMN analysis due to noisy data, yielding 14 analyzed for symptoms/cognition and 11 per group for MMN. There were no significant demographic differences between individuals who received D-serine vs. placebo first in the double-blind study (Table 1). 2.1.2. Open-label/bitopertin design differences Open-label subjects in the present report only include subjects from the NKI site, as biomarker assessment was not done at the other sites. Subjects received doses of 30 (5 subjects), 60 (8 subjects) or 120 mg/kg/d (6 subjects) for four weeks. Antipsychotic doses were limited to ≤1500 mg/d chlorpromazine (CPZ) equivalents (Woods, 2003), but otherwise identical criteria to the double-blind report were used. Table 1 Baseline demographics and outcome measures.

Demographics

a

Age Male (%) In-patient (%) Age of 1st treatment Duration of illness Schizoaffective (%) CPZ equivalents

DBa crossover (n = 14)

OLa (n = 19)

40 ± 11 93% 71% 16 ± 8 23 ± 12 14% 965 ± 760

44 ± 9 95% 100% 20 ± 6 25 ± 7 5% 589 ± 220

Mean ± SD or %, DB: double-blind and OL: open-label.

3

In the bitopertin study, subjects were randomized in a 3:2 ratio to bitopertin (10 mg) versus placebo for six-weeks. There were no PANSS inclusion criteria, and functional impairment was determined by a GAF score between 31 and 50 and a CGI-S score of ≥ 4. Subjects were excluded for abnormal hemoglobin, and for Abnormal Involuntary Movement Scale global severity score N3 (moderate) and/or Facial and Oral Movement items with a score N2 (mild). 2.2. MMN MMN was assessed at baseline and at the end of treatment or phase, collected in a single session per day. As per (Friedman et al., 2012), stimuli consisted of a sequence of four complex tones (one standard tone and duration, intensity, and frequency deviants) presented in random order at stimulus onset asynchronies of 500–505 msec. The standard stimulus (70% sequential probability) was a harmonic tone composed of three superimposed sinusoids (500, 1000, and 1500 Hz), and was 100 msec in duration, approximately 85 dB, and with 5-msec rise and fall time. Duration, pitch, and intensity deviants (10% probability each) were 150 msec in length, 10% lower in pitch, and of 10 dB lower in intensity, respectively, relative to standards, each presented with sequential probability of 10%. Continuous electroencephalographic data were collected, along with trigger timing information using the Cognitrace system developed by ANT Neuro (ANT Neuro, Enschede, the Netherlands), maintaining an impedance b10 Ω, with a sample rate of 512 Hz, from 65 electrode positions distributed over the scalp. ERP analysis was performed offline using MATLAB (Mathworks, Natick, MA) and the open source EEG analysis toolbox, Fieldtrip (http://www.ru.nl/fcdonders/fieldtrip/). A 0.1 to 100 Hz filter was applied, as well as a 60 Hz line noise filter, and the electrode information was re-referenced to the average linked mastoids for auditory stimuli. Epochs were defined from 1500 ms before the stimulus to 2000 ms afterwards. Artifact rejection was performed manually with Fieldtrip's visual rejection tool, (which plots epochs and channels based on the average absolute value of their amplitude), and removing trials that exceeded 150 μV. Epochs were then averaged for each stimulus separately to generate ERPs and low pass filtered at 30 Hz. 2.2.1. ERP The three MMN ERP conditions, frequency, intensity and duration, were created by taking the difference between the respective deviant stimuli average waveforms, and the standard stimulus average waveform. As previously (Friedman et al., 2012), amplitudes were chosen from the fronto-central region (Fz). As previously (Kantrowitz et al., 2016), mean amplitudes were calculated across 100–250 msec for duration and 50–200 msec for intensity and frequency. For the standard stimuli ERP, peaks were analyzed within 70–130 ms for N1. There were no significant baseline final or between treatment differences in accepted trials (sweep count: Supplemental Table 1). 2.2.2. Time-frequency Time-frequency analysis of MMN were constructed using two complementary approaches in the double-blind D-serine study only. Analysis windows and bands were identical to our prior study (Lee et al., 2017). For evoked analyses, ERP waves were transformed by multitaper method with Hanning window implemented with Fieldtrip Open Toolbox with 10 ms time resolution and 1 Hz step of frequency resolution (Oostenveld et al., 2011). For single-trial analyses, ITC and baselinecorrected single-trial power were obtained from BESA 5.1 using a complex demodulation procedure with 2 Hz frequency resolution and 25 ms time resolution. For standard stimuli, θ-frequency was analyzed across 25–275 ms and α-frequency across 25–125 msec. For MMN, θ and αfrequency were analyzed across 50–300 msec for duration and across 25–275 msec for frequency and intensity.

Please cite this article as: Kantrowitz, J.T., et al., Improvement in mismatch negativity generation during d-serine treatment in schizophrenia: Correlation with symptoms, Schizophr. Res. (2017), http://dx.doi.org/10.1016/j.schres.2017.02.027

4

J.T. Kantrowitz et al. / Schizophrenia Research xxx (2017) xxx–xxx

2.3. Symptom and cognitive assessments Symptom assessments were performed biweekly using the PANSS total and five-factor scores (Lindenmayer et al., 1994). The MCCB (minus the social cognition domain) was used for neurocognitive assessment, administered before and after each treatment phase. Normalized T-scores were calculated from the six tested domains based on age and education level using the MCCB norms (Nuechterlein and Green, 2006). The tone-matching task (Kantrowitz et al., 2014), was assessed in the double-blind study along with the MCCB in a subset (n = 11). 2.4. Laboratory/safety examination Primary safety focus in D-serine studies was on renal function (serum creatinine, BUN, microscopic urinalysis). Additional safety measures included ECG, liver function tests (LFT), CBC and general chemistry. All safety measures were conducted biweekly. Baseline and steady state pharmacokinetics (PK) was conducted (Kantrowitz et al., 2010). 2.5. Statistical analysis and meta-analysis

groups standard deviations of the pre-treatment severity scores. We tested for publication bias with the Rank Correlation Test and provide a Funnel Plot for Asymmetry. The heterogeneity of the treatment effect is measured by the heterogeneity parameter τ2 from the random effects models, and we tested whether the sample size is related to the heterogeneity. 3. Results 3.1. Mismatch negativity Voltage-topography distribution was similar to, but visibly reduced compared to controls (Fig. 1, bottom) with strong test-retest reliability across sessions and studies (Cronbach α: MMN = 0.75). Primary MMN analysis was conducted for mean amplitude (ERP) in the double-blind study, with a secondary analysis conducted in time-frequency. 3.1.1. MMN ERP (Fig. 1, Table 2) In the double-blind cross-over, D-serine treatment led to a significant improvement vs. placebo for frequency MMN (F1,9.1 = 26.0, p = 0.001, d = 2.3), without a treatment-by-order effect (e.g. D-serine first vs. placebo first) for baseline (F1,9.4 = 0.0, p = 0.99) or final (F1,8.1 = 0.2, p = 0.67) values. No significant differences were seen across deviants (F1,51.0 = 2.0, p = 0.16, d = 0.35), nor individually for MMN to either duration (F1,19 = 0.006, p = 0.94, d = −0.05) or intensity (F1,7.7 = 0.14, p = 0.72, d = − 0.17) deviants. Significance of improvement in frequency MMN remained even with Bonferroni correction applied across the 3-deviant types (adjusted p value for significance of p = 0.017). As expected (Lee et al., 2017), no significant differences were seen in ERP for the standard tone for N1 (F1,11.9 = 2.9, p = 0.12, d = 0.76).

The primary analyses were conducted in the double-blind D-serine, in accordance with clinicaltrials.gov. A secondary, separate analysis (Section 3.8), incorporated the open label D-serine and bitopertin studies, which evaluated the treatment effects across studies using the pooled-placebo group in order to minimize placebo contributions to cross-study results. Subjects were compared between randomized groups within study with respect to baseline characteristics using χ2 test for categorical and 2-samples t-test for continuous variables. The treatment effects were estimated with respect to symptoms (PANSS), neurophysiological measures (MMN and time-frequency) and cognition (MCCB). Linear mixed effects models (Diggle et al., 2002) were used to accommodate correlated responses of subjects in the crossover double-blind D-serine study, and included indicators for treatment (D-serine or placebo), and adjusted for the baseline levels of the outcomes. Since PANSS was measured repeatedly over the course of treatment, we included time by treatment interactions. In addition to random subject intercepts, these mixed effects models included random subject slopes to account for the correlation between the repeated bi-weekly assessments of subjects' symptoms. Between-group effectsizes were calculated from F-values using Cohen's d expressed in SD units. Values in the text are mean ± SD unless otherwise specified. The associations between potential biomarkers (MMN ERP and time frequency) and clinical outcomes (PANSS/MCCB) was studied based on Pearson correlations and linear regression within the double-blind Dserine study, controlling for treatment-type and deviant-type. The relationships were characterized with respect to the F-test for significance of a multivariate regressions model. Since the lowest possible PANSS score on an individual item is 1, not 0, the percent change in PANSS was calculated by (baseline-final)/(baseline-number of items in a factor).

3.1.2.1. Standards. D-Serine treatment led to a significant, large effect size improvement vs. placebo in evoked power for the α band (t10 = 2.4, p = 0.036, d = 0.81, Fig. 2, left) in an exploratory paired t-test, appearing to normalize deficits in evoked α power relative to previous findings with controls (Lee et al., 2017) (Fig. 2, upper right). Trend level significance was retained controlling for baseline differences between groups (F1,11.5 = 3.3, p = 0.096) in a mixed model. Evoked α power significantly correlated with both frequency (r = − 0.49, p = 0.023) and duration (r = −0.47, p = 0.03) ERP MMN amplitude. Correlations remained significant after control for treatment-type for both frequency (partial r = −0.55, p = 0.012, Fig. 2, middle right) and duration (partial r = −0.47, p = 0.036), suggesting similar mechanisms of evoked α power for standards and MMN ERP.

2.5.1. Meta-analysis The analysis was conducted using Restricted Maximum Likelihood (REML) estimation as implemented in the function rma of the library metaphor in R (R Core Team; Viechtbauer, 2010). Using a random effects model, we evaluated the effect of adjunctive D-serine (along with the related compound D-alanine) on negative symptoms by combining findings from six studies (Heresco-Levy et al., 2005; Lane et al., 2010; Tsai et al., 1998; Tsai et al., 2006; Weiser et al., 2012), including the double-blind subjects in present report. Studies using open-label D-serine, acutely ill subjects (Ermilov et al., 2013; Lane et al., 2010), monotherapy or treatment combined with other interventions were excluded. Effect sizes (Cohen's d) were computed as the difference between drug and placebo with respect to change scores: pre- minus post-treatment symptoms severity scores, divided by the pooled across treatment

3.1.2.2. MMN. D-Serine treatment led to a significant reduction vs. placebo in evoked power for the α band across deviants (F1,48.4 = 4.4, p = 0.042, d = 0.52), also appearing to normalize elevated evoked α power relative to previous findings with controls (Lee et al., 2017). Significant between group differences were also seen for single trial power in the θ band (F1,51.1 = 6.0, p = 0.018, d = 0.61) across deviants. Individual significance was seen for both MMN to frequency (F1,19 = 5.8, p = 0.027, d = 1.1) and duration (F1,9.0 = 5.3, p = 0.047, d = 1.0) for evoked power for the α band and for MMN to frequency (F1,19 = 9.8, p = 0.006, d = 1.4) for single trial θ power. Non-significant moderate effect size increases were seen for ITC. Across deviants, single trial (r = 0.31, p = 0.014) power the θ band for MMN significantly correlated with ERP MMN amplitude, but evoked power for the α band (r = − 0.14, p = 0.28). Correlations remained

3.1.2. Time frequency MMN Time-frequency analysis was conducted in the double-blind study for MMN (Fig. 2, Table 3) as a secondary analysis. As previously (Lee et al., 2017), we analyzed both evoked power and single trial analyses (separate single-trial power and ITC).

Please cite this article as: Kantrowitz, J.T., et al., Improvement in mismatch negativity generation during d-serine treatment in schizophrenia: Correlation with symptoms, Schizophr. Res. (2017), http://dx.doi.org/10.1016/j.schres.2017.02.027

J.T. Kantrowitz et al. / Schizophrenia Research xxx (2017) xxx–xxx

5

Fig. 1. Voltage topography maps for double blind D-serine and placebo (bottom) subjects for mismatch negativity (MMN) to frequency, shown at peak latencies. Analyzed electrode noted by blue circle (Fz). Bottom left: MMN to frequency in controls, modified from (Friedman et al., 2012) and shown for illustration purposes and was not analyzed for this study. Right: Grand average waveforms for baseline (dashed) and final (solid) for double-blind D-serine (top right in red) and placebo (bottom right in blue). Box highlights analysis window for mean amplitude.

Table 2 Primary outcomes: baseline and final scores for D-serine studies (mean ± sd). Treatment type

Active double blind only

Active double blind + open label

Placebo

Baseline

Baseline

Baseline

Final

Final

Double blind Final

Double blind + open label

Statistics (F/p) Effect size (d) Statistics (F/p) Effect size (d)

MMN (μV)a MMN_frequency MMN_duration MMN_intensity

−0.71 ± 1.0 −1.21 ± 0.6 −0.46 ± 1.1 −0.84 ± 0.7 −0.79 ± 1.3 −0.21 ± 0.9 26/0.001 −1.00 ± 1.0 −1.03 ± 1.4 −0.81 ± 1.2 −0.70 ± 1.1 −0.92 ± 1.1 −0.97 ± 0.9 0.01/0.94 −0.82 ± 0.9 −0.36 ± 0.9 −0.96 ± 1.0 −0.63 ± 0.9 −1.08 ± 1.1 −0.73 ± 1.3 0.1/0.72

2.3 −0.05 −0.17

PANSSa Total Negative Positive Cognitive Depression Excitement

82.2 ± 7.9 20.8 ± 4.1 13.1 ± 4.2 12.1 ± 3.4 12.2 ± 3.9 7.6 ± 3.5

77.9 ± 9.3 19.4 ± 3.4 13.1 ± 4.1 11.1 ± 3.8 11.4 ± 3.5 7.8 ± 2.8

26.0 ± 7.5 15.9 ± 12.9

26.2 ± 7.6 15.8 ± 14.6

27.0 ± 7.8 16.2 ± 13.1

24.1 ± 10.0

25.6 ± 10.4

21.0 29.3 34.6 32.6

21.4 30.6 33.6 31.5

MCCBa Total Speed of processing Attention & vigilance Working memory Verbal learning Visual learning Reasoning & problem solving

± ± ± ±

12.3 5.1 13.7 8.2

± ± ± ±

12.7 5.4 12.3 7.0

19.1/0.001 0.5/0.48 0.0/0.87

1.6 −0.26 0.0

79.3 ± 9.0 21.0 ± 2.2 13.2 ± 4.5 11.1 ± 3.4 11.3 ± 3.8 7.3 ± 2.4

80.0 ± 10.3 20.9 ± 2.1 13.2 ± 4.1 10.6 ± 4.2 11.9 ± 3.5 8.1 ± 3.9

5.4/0.02 0.85/0.36 0.06/0.82 0.6/0.43 1.4/0.24 0.5/0.47

0.8 0.88 −0.04 0.32 0.69 0.19

28.4 ± 7.8 18.8 ± 12.6

26.9 ± 7.6 15.8 ± 12.9

25.7 ± 7.9 15.7 ± 15.4

1.1/0.31 0.0/0.99

0.41 0

4.2/0.052 0.3/0.58

0.67 0.18

26.1 ± 12.7

28.1 ± 11.4

25.8 ± 11.3

24.9 ± 9.7

2.6/0.13

0.63

3.6/0.08

0.62

23.1 29.3 33.5 34.3

23.6 31.4 34.2 34.8

21.7 29.7 36.9 32.5

20.9 29.4 32.9 31.5

0.2/0.64 3.0/0.12 1.1/0.3 0.02/0.89

0.18 0.68 0.41 −0.06

0.3/0.57 2.6/0.14 1.7/0.21 0.0/0.98

0.18 0.53 0.43 0

± ± ± ±

13.2 5.3 12.8 7.3

± ± ± ±

13.2 5.9 11.6 8.3

± ± ± ±

12.0 4.0 13.8 8.7

± ± ± ±

13.1 6.3 15.1 7.5

Bold values indicate significance at p b 0.05. Italic values indicate significance at p b 0.01. a n = 14 for D-serine double blind, n = 14 for placebo and n = 33 for D-serine double blind + open label and for MCCB and PANSS. 3 active and 3 D-serine placebo subjects were dropped from MMN due to noisy data. Statistics for PANSS are for treatment by time analysis effect size calculated from baseline final treatment effects.

Please cite this article as: Kantrowitz, J.T., et al., Improvement in mismatch negativity generation during d-serine treatment in schizophrenia: Correlation with symptoms, Schizophr. Res. (2017), http://dx.doi.org/10.1016/j.schres.2017.02.027

6

J.T. Kantrowitz et al. / Schizophrenia Research xxx (2017) xxx–xxx

Fig. 2. Left: Time frequency plots for baseline final evoked power in response to the standard tone. White * notes the analyzed area (α band 7–12 Hz, 25–125 msec). Upper right: Evoked α power in controls, modified from (Lee et al., 2017) and shown for illustration purposes and was not analyzed for this study. Middle right: Scatter plot of Mismatch negativity (MMN) to frequency vs. Evoked α power to the standard tone. Evoked α power was log transformed for figure. Partial r, controlling for treatment type in text. Bottom right: Scatter plot of baselinefinal change in the Positive and Negative Symptom Scale (PANSS) Total vs. Final MMN to frequency. D-Serine (red) and placebo (blue) for all figures.

significant after control for deviant-type and treatment-type for single trial power (partial r = 0.25, p = 0.05), suggesting similar mechanisms between time frequency analysis and MMN ERP.

3.2. PANSS A significant treatment by time interaction was observed for total PANSS symptoms in the double-blind study (F1,68.6 = 5.4, p = 0.023, d = 0.80, Table 2), reflecting significant 8.1 ± 13.9% improvement in the D-serine group (t13 = 5.0, p = 0.044) with no significant change during placebo (t13 = 0.2, p = 0.7). The treatment-by-order effect was non-significant for both for baseline (F1,12 = 3.2, p = 0.1) and final (F2,19.7 = 0.4, p = 0.66) PANSS total scores. Although no significant treatment by time interaction was observed for PANSS negative symptoms (F1,81.3 = 0.9, p = 0.36, Table 2), a significant treatment effect was observed at study end after controlling for baseline scores in the double-blind study (F1,10.8 = 5.0, p = 0.047, d = 0.88). Similar to total PANSS, a significant within group improvement in negative symptoms was seen in the D-serine group (t13 = 2.5, p = 0.028, 8.2 ± 13.8%) with no significant change during placebo (t13 = 0.1, p = 0.9). No significant treatment by time or treatment effects were seen in other PANSS factors.

3.3. Cognition Baseline MCCB scores were ~ 2·5 sd below normative levels, suggesting clinically significant cognitive deficits at baseline (Table 2). The MCCB composite change was not significant in the double-blind subjects (F1,11.9 = 1.1, p = 0.31, d = 0.41). Non-significant, moderate effect size improvements were seen in both the Verbal Memory and Attention and Vigilance domains (Table 2) within the double-blind. In addition to MCCB, a subset of patients in the double-blind study was assessed on behavioral auditory measures. On the tone-matching task, a trend towards improvement was seen during D-serine (F1,10 = 4.4, p = 0.06, d = 0.67), but not during placebo (F1,9 = 0.8, p = 0.41, d = 0.30) treatment. However, the between-treatments difference was not significant (F1,19 = 0.7, p = 0.41, d = 0.37).

3.4. Relationship between outcomes Within the double-blind D-serine study, final frequency MMN amplitude predicted change in total PANSS (r = 0.5, p = 0.018), which remained significant after control for treatment-type (F2,19 = 8.0; p = 0.001, R2 = 0.46, partial r = 0.63, p = 0.002, Fig. 2, bottom right). Similarly, and consistent with our previous findings (Lee et al.,

Table 3 Final scores (mean ± sd) for standard ERP and time-frequency measures for MMN and standards.

ERP (Standard) Time Frequency (Standard) θ: 25-275 msec α: 25-125 msec

Time Frequency (MMN) Mean of θ or α band Across MMN to duration: 50-300 msec, frequency and intensity: 25~275 msec

D-serine

Placebo

Statistics (F/p)

Effect size (d)

N1, Standard tone

-0.20 ± 0.39

-0.23 ± 0.31

2.9/0.16

0.76

Evoked θ Single trial θ ITC θ Evoked α Single trial α ITC α Evoked θ Evoked α Single trial θ

0.02 ± 0.03 -0.02 ± 0.02 0.08 ± 0.05 0.04 ± 0.05 -0.02 ± 0.02 0.06 ± 0.05 0.01 ± 0.01 0.02 ± 0.02 -0.03 ± 0.08

0.02 ± 0.02 -0.02 ± 0.02 0.09 ± 0.04 0.03 ± 0.03 -0.02 ± 0.01 0.09 ± 0.06 0.01 ± 0.01 0.04 ± 0.07 0.03 ± 0.1

0.01/0.93 0.6/0.48 0.3/0.61 3.3/0.096 0.6/0.50 0.3/0.59 0.2/0.64 4.4/0.042 6.0/0.018

-0.14 0.35 -0.25 0.81 0.35 0.25 -0.11 0.52 0.61

ITC θ

0.02 ± 0.04

0.01 ± 0.01

2.4/0.13

0.39

Please cite this article as: Kantrowitz, J.T., et al., Improvement in mismatch negativity generation during d-serine treatment in schizophrenia: Correlation with symptoms, Schizophr. Res. (2017), http://dx.doi.org/10.1016/j.schres.2017.02.027

J.T. Kantrowitz et al. / Schizophrenia Research xxx (2017) xxx–xxx

2017), final ITC across MMN deviants predicted change in total PANSS (r = − 0.28, p = 0.02), which remained significant after control for treatment-type and deviant-type (F2,62 = 5.3; p = 0.03, R2 = 0.2, partial r = −0.25, p = 0.05). Correlations between the MCCB and MMN amplitude were not significant within the double-blind study. 3.5. PK/PD Steady-state measurements of serum D-serine in double-blind (139.3 ± 82 μM) were statistically similar to steady-state measurements (137.5 ± 30 μM) reported previously (Kantrowitz et al., 2010) (t20 = 0.06, p = 0.95). As expected, all subjects returned to baseline D-serine

levels after the two-week washout.

3.6. Meta-analysis The meta-analysis of studies of D-serine and related compounds showed a highly significant improvement in negative symptoms with a medium overall effect size (d = 0.70, p = 0.006), after co-varying for sample-size (Fig. 3A/B). The model indicated that the treatment effect is heterogeneous (heterogeneity τ2 = 0.34, p b 0.001), but without significant effect of sample size (p = 0.16). Although the heterogeneity of the treatment effect did not disappear following covariation for sample size, its magnitude and significance were reduced (heterogeneity τ2 = 0.26, p = 0.015). 3.7. Safety measures In the double-blind D-serine study, no clinically significant side effects were observed and all abnormal renal measures seen in the context of placebo treatment, including trace blood and transiently elevated creatinine. There were no drug-related effects on other safety parameters. 3.8. Contrast with open-label D-serine and bitopertin study 3.8.1. Open-label D-serine results (Table 2) Data was from the 21 subjects at the NKI site (Kantrowitz et al., 2010). Two subjects were withdrawn because of non-serious adverse events/withdrawn consent prior to post baseline outcome assessments, yielding 19 analyzed subjects. There were no significant demographic differences between those participating in the double-blind vs. openlabel trials (Table 1). Frequency MMN (F1,11.7 = 19.1, p = 0.001, d = 1.6) retained significance, while the magnitude of change in PANSS was statistically similar

7

in double-blind and open-label subjects (F1,30 = 0.3, p = 0.62). Trend level improvement was seen in the MCCB total across all D-serine patients (F1,22.6 = 4.4, p = 0.052, d = 0.67). 3.8.2. Bitopertin results As previously reported (Da Costa et al., 2015; Kantrowitz et al., submitted for publication), 29 subjects were enrolled (bitopertin: 17 and placebo: 12), without demographic differences between bitopertin and placebo groups. Two patients assigned to bitopertin were withdrawn: one for pain from a pre-existing hernia, and one for abnormal LFT's possibly related to study treatment. Reliability of MMN was similar across studies. No significant treatment-by-deviant effect was seen for MMN for frequency (F1,24 = 0.01, p = 0.94, d = 0), the PANSS total (F1,85 = 0.2, p = 0.68, d = −0.18) or any other outcome. In an exploratory, post hoc analysis across all 63 subjects from the three studies, D-serine retained significance vs. the pooled placebo group for frequency MMN (d = 0.71, p = 0.001) and total PANSS (d = 0.61, p = 0.005), and vs. bitopertin for MMN to frequency deviants (d = 0.41, p = 0.01). 4. Discussion The present paper reports on the effects of D-serine on NMDARbased biomarkers in schizophrenia. Consistent with the current and prior meta-analysis (Cho et al., 2016; Tsai and Lin, 2010), D-serine treatment led to a significant, moderate-large effect-size reduction in total and negative symptoms in stabilized patients. Additionally, we observed significant biomarker modulation with D-serine on MMN, a measure known to be sensitive to NMDAR dysfunction in animal models. MMN, moreover, was correlated with improvement in symptoms, suggesting a potential link between functional target engagement and clinical response. In the present study, MMN showed high testretest reliability, supporting their use as target engagement measures (Javitt et al., 2011). Our findings suggest first that D-serine treatment is associated with significant objective alteration in brain function, and, second, that the measures chosen for study, particularly auditory MMN, may serve as effective target engagement biomarkers for future NMDAR-related treatment development. 4.1. Comparison with prior trials of NMDAR agonists To date, D-serine has primarily been studied at a dose of 30 mg/kg/d with a recent meta-analysis suggesting significant but small-moderate improvement in negative and positive symptoms (Cho et al., 2016). We have recently demonstrated significant moderate effect size improvement in negative symptoms (d = 0.68) during continuous

Fig. 3. (A) Forest plot: the solid squares indicate the observed effect sizes and their size is proportional to the inverse of their variance; the whiskers extend to the 95% confidence intervals; the red rhomboids indicates the model-based estimates of the effect sizes (adjusting for total sample size). (B) Funnel plot for assessing evidence for publication bias: the residuals from model with total sample size as covariate. There was no indication for publication bias as indicated by the Rank Correlation Test for Funnel Plot Asymmetry (Kendall's τ = 0.60, p = 0.14).

Please cite this article as: Kantrowitz, J.T., et al., Improvement in mismatch negativity generation during d-serine treatment in schizophrenia: Correlation with symptoms, Schizophr. Res. (2017), http://dx.doi.org/10.1016/j.schres.2017.02.027

8

J.T. Kantrowitz et al. / Schizophrenia Research xxx (2017) xxx–xxx

treatment with D-serine 60 mg/kg/d in clinically high risk population (Kantrowitz et al., 2015b), as well as significant, intercorrelated improvements in auditory plasticity, θ and β-frequency response and MMN after repeated intermittent (once-weekly) treatment administration of D-serine 60 mg/kg/d (Kantrowitz et al., 2016). The present report is the first study to evaluate effects of continuous (daily) D-serine (60 mg/kg/d) treatment simultaneously on neurophysiological biomarkers and symptoms in schizophrenia in a double-blind, placebo controlled trial, demonstrating similar results to the open-label study. Our updated meta-analysis including the present report continues to demonstrate significant clinical effect with moderate-large (d = 0.7) effect size (Fig. 3A/B). Similar results were seen across double-blind and open-label trials. As opposed to D -serine (and the closely related compound D alanine) which have shown consistent positive results in schizophrenia, other approaches for modulation of NMDAR, including selective GlyT1 inhibitors such as bitopertin have not been successful and have been associated with a very narrow therapeutic window (Section 3.8). By contrast to D -serine results, bitopertin did not significantly affect either symptoms or MMN at the dose tested (10 mg). The lack of change of MMN in the bitopertin study is consistent with lack of clinical effect, and thus suggests that MMN may have negative, as well as positive, predictive value in predicting efficacy of novel compounds. By contrast, sarcosine, a mixed GlyT1/System A transport antagonist has been reported to have robust, clozapine-like effects on clinical symptoms (Lane et al., 2005; Lane et al., 2010), but has not yet been studied using neurophysiological measures. Similarly, benzoate, a DAAO inhibitor, which reduces break down of D-serine, has shown pro-cognitive benefits alone and when combined with sarcosine (Lane et al., 2013; Lin et al., 2015). Future studies comparing high dose D-serine vs. sarcosine or benzoate effects on biomarkers and symptoms may therefore be warranted. 4.2. Biomarkers Similar to our trial of once-weekly D-serine (Kantrowitz et al., 2016), frequency MMN showed both greatest sensitivity to D-serine effects and correlated highly with change in symptoms, further supporting MMN's link to underlying pathophysiology. This relationship between symptoms and MMN suggests that before large scale clinical trials are conducted, MMN could be used in early stage drug development programs of novel NMDAR modulating treatments to ensure D-serine-like target engagement. In addition to D-serine, frequency MMN has shown sensitivity to effects of N-acetylcysteine, a compound that modulates NMDAR indirectly (Lavoie et al., 2008). In a recent mechanistic analysis of MMN generation, we have shown that frequency MMN reflects primarily cortical change, whereas both intensity and duration MMN involve both cortical and subcortical components (Lee et al., 2017). Furthermore, as compared to deficits in duration MMN that tend to predominate early in the course of the illness, deficits in frequency MMN tend to predominate later (Haigh et al., 2017; Todd et al., 2008) and are more tied to neurodegeneration (Friedman et al., 2012). Thus, differential response to frequency vs. duration/intensity MMN may reflect differential underlying pathophysiological mechanisms. Consistent with our prior work (Kantrowitz et al., 2016), we demonstrated significant moderate effect size changes in time-frequency measures. Consistent with our previous findings with controls (Lee et al., 2017) (Fig. 2, upper right), D-serine appeared to normalize evoked α power, leading to reduced activation in response to standard tones and reduced suppression for MMN. Additionally, as opposed to ERP, in which significant changes were on seen during frequency MMN, significant evoked power changes were seen for standards and both frequency and duration MMN, suggesting increased sensitivity to change for time frequency measures.

4.3. Limitations Despite the significant therapeutic effects, some limitations must be considered. First, the improvement in symptoms, while significant, was nevertheless smaller than the 15–20% reduction observed in prior D -serine studies with 30 mg/kg/day (Heresco-Levy et al., 2005; Tsai et al., 1998). Although this pattern of result could represent an inverted U-shaped dose-response relationship, we feel that a more likely explanation relates to the chronicity of the sample and the relatively high doses of antipsychotic medication vs. previous studies. Second, time frequency analysis was exploratory, and limited by small sample size and between group baseline differences. Third, although the D -serine and the bitopertin studies were conducted by the same investigators and using the same equipment and neurophysiological paradigm, there were some differences in inclusion criteria (Section 2.1.2), and the across study comparisons should be considered exploratory. 4.4. Conclusions In summary, NMDAR agonists remain under clinical development, but ideal agents and doses remain to be determined. The present study replicates significant effects of D-serine on clinical symptoms and objective neurophysiological markers of brain dysfunction in schizophrenia. Furthermore, improvement on these markers correlates significantly with clinical response. In early-stage clinical development, determining the ideal dose to enter into definitive studies remains a critical challenge. The present study suggests that neurophysiological markers in general and auditory MMN in particular may serve as useful target engagement biomarkers for early stage clinical drug development, and argue that NMDAR agents differ on functional target engagement, potentially explaining the varied findings of efficacy. MMN may be useful to provide Go-No go signals for early stage studies. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.schres.2017.02.027. Role of the funding source D-Serine studies were supported by the NIH grants (U01 MH074356 and P50 MH086385) to DCJ. The bitopertin study was supported by an investigator initiated grant from F. Hoffmann-La Roche to DCJ. The NIH and F. Hoffmann-La Roche had no role in the final analysis or decision to submit for publication.

Contributors Dr. Kantrowitz, Ms. Silipo and Dr. Javitt designed the study and wrote the protocol. Ms. Lehrfeld, Dr. Nolan, Ms. Silipo and Dr. Shope managed the clinical and cognitive outcomes and analyses. Dr. Kantrowitz, Dr. Javitt, Mr. Epstein and Dr. Lee conducted the neurophysiological analysis. Dr. Kantrowitz and Petkova undertook the statistical analysis, and Dr. Kantrowitz wrote the first draft of the manuscript. All authors contributed to and have approved the final manuscript. Conflict of interest Dr. Kantrowitz reports having received consulting payments within the last 36 months from Vindico Medical Education, Annenberg Center for Health Sciences at Eisenhower, Health Advances, LLC, SlingShot, Strategic Edge Communications, Havas Life and Cowen and Company. He has conducted clinical research supported by the NIMH, Merck, the Stanley Foundation, Roche-Genetech, Forum, Sunovion, Novartis, Lundback, Alkermes, NeuroRx, Pfizer and Lilly. He owns a small number of shares of common stock in GlaxoSmithKline. Dr. Javitt reports having received consulting payments within the last 36 months from Sunovion, Forum, and Takeda. He has received research support from Roche. He holds intellectual property rights for use of NMDA modulators in treatment of neuropsychiatric disorders. He holds equity in Glytech, AASI, and NeuroRx, and serves on the advisory board of Promentis and NeuroRx. All other co-authors report no conflicts. Acknowledgements Clinical and cognitive data from NCT00322023 was previously published in Kantrowitz JT, et al. High dose D-serine in the treatment of schizophrenia. Schizophrenia Research. 2010; 121(1–3):125–30. Data has been presented in part at the Society for Biological Psychiatry 2014 annual meeting and the NCDEU 2015 annual meeting. Full trial protocols are available upon request.

Please cite this article as: Kantrowitz, J.T., et al., Improvement in mismatch negativity generation during d-serine treatment in schizophrenia: Correlation with symptoms, Schizophr. Res. (2017), http://dx.doi.org/10.1016/j.schres.2017.02.027

J.T. Kantrowitz et al. / Schizophrenia Research xxx (2017) xxx–xxx

References Addington, D., Addington, J., Maticka-Tyndale, E., 1994. Specificity of the Calgary Depression Scale for Schizophrenics. Schizophr. Res. 11 (3), 239–244. Alberati, D., Moreau, J.L., Lengyel, J., Hauser, N., Mory, R., Borroni, E., Pinard, E., Knoflach, F., Schlotterbeck, G., Hainzl, D., Wettstein, J.G., 2012. Glycine reuptake inhibitor RG1678: a pharmacologic characterization of an investigational agent for the treatment of schizophrenia. Neuropharmacology 62 (2), 1152–1161. Arango, C., Nasrallah, A.T., Lawrie, S., Lohmann, T.O., Zhu, J.L., Garibaldi, G., Wang, A., Bugarski-Kirola, D., 2014. Efficacy and safety of adjunctive bitopertin (5 and 10 mg) versus placebo in subjects with persistent predominant negative symptoms of schizophrenia treated with antipsychotics—results from the phase III DayLyte study. Schizophr. Res. 158 (e1). Balu, D.T., Coyle, J.T., 2015. The NMDA receptor ‘glycine modulatory site’ in schizophrenia: D-serine, glycine, and beyond. Curr. Opin. Pharmacol. 20, 109–115. Blaettler, T., Bugarski-Kirola, D., Fleischhacker, W.W., Bressan, R., Arango, C., Abi-Sabb, D., Napieralski, J., Dixon, M., Marder, S., 2014. Efficacy and safety of adjunctive bitopertin (10 and 20 mg) versus placebo in subjects with persistent predominant negative symptoms of schizophrenia treated with antipsychotics—results from the Phase III FlashLyte Study. Schizophr. Res. 158, e2–e3. Buchanan, R.W., Javitt, D.C., Marder, S.R., Schooler, N.R., Gold, J.M., McMahon, R.P., Heresco-Levy, U., Carpenter, W.T., 2007. The Cognitive and Negative Symptoms in Schizophrenia Trial (CONSIST): the efficacy of glutamatergic agents for negative symptoms and cognitive impairments. Am. J. Psychiatry 164 (10), 1593–1602. Bugarski-Kirola, D., Iwata, N., Sameljak, S., Reid, C., B., T., Millar, L., Marques, T.R., Garibaldi, G., Kapur, S., 2016. Efficacy and safety of adjunctive bitopertin versus placebo in patients with suboptimally controlled symptoms of schizophrenia treated with antipsychotics—results from the SearchLyte clinical trial programme. Lancet Psychiatry 3 (12), 1115–1128. Cho, S.E., Na, K.S., Cho, S.J., Kang, S.G., 2016. Low D-serine levels in schizophrenia: a systematic review and meta-analysis. Neurosci. Lett. 634, 42–51. Coyle, J.T., 2012. NMDA receptor and schizophrenia: a brief history. Schizophr. Bull. 38 (5), 920–926. Da Costa, D., Umbricht, D., Youssef, E., Sun, S., Javitt, D.C., Fontoura, P., Santarelli, L., 2015. Effects of Bitopertin on Biomarker Measures of Cognitive Dysfunction in Adults with Schizophrenia and Schizoaffective Disorder: Results From a Ph1b, Double Blind, Placebo Controlled Parallel Arm Study. American College of Clinical Psychopharmacology, Miami Beach, FL. Diggle, P.J., Heagerty, P., Liang, K.Y., Zeger, S.L., 2002. Analysis of Longitudinal Data. 2 ed. Oxford University Press, Oxford, UK. Erickson, M.A., Ruffle, A., Gold, J.M., 2016. A meta-analysis of mismatch negativity in schizophrenia: from clinical risk to disease specificity and progression. Biol. Psychiatry 79 (12), 980–987. Ermilov, M., Gelfin, E., Levin, R., Lichtenberg, P., Hashimoto, K., Javitt, D.C., Heresco-Levy, U., 2013. A pilot double-blind comparison of D-serine and high-dose olanzapine in treatment-resistant patients with schizophrenia. Schizophr. Res. 150 (2–3), 604–605. First, M.B., Spitzer, R.L., Gibbon, M., Williams, J., 1997. Structural Clinical Interview for DSM-IV Axis I Disorders (SCID-IV). Biometrics Research Department, New York State Psychiatric Institute, New York. Friedman, T., Sehatpour, P., Dias, E., Perrin, M., Javitt, D.C., 2012. Differential relationships of mismatch negativity and visual P1 deficits to premorbid characteristics and functional outcome in schizophrenia. Biol. Psychiatry 71 (15), 521–529. Friston, K., 2005. A theory of cortical responses. Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci. 360 (1456), 815–836. Garrido, M.I., Kilner, J.M., Stephan, K.E., Friston, K.J., 2009. The mismatch negativity: a review of underlying mechanisms. Clin. Neurophysiol. 120 (3), 453–463. Green, M.F., Hellemann, G., Horan, W.P., Lee, J., Wynn, J.K., 2012. From perception to functional outcome in schizophrenia: modeling the role of ability and motivation. Arch. Gen. Psychiatry 1–9. Gunduz-Bruce, H., Reinhart, R.M., Roach, B.J., Gueorguieva, R., Oliver, S., D'Souza, D.C., Ford, J.M., Krystal, J.H., Mathalon, D.H., 2012. Glutamatergic modulation of auditory information processing in the human brain. Biol. Psychiatry 71 (11), 969–977. Guy, W., 1976. ECDEU Assessment Manual for Psychopharmacology: Revised. National Institute for Mental Health, Rockville, MD. Haegens, S., Barczak, A., Musacchia, G., Lipton, M.L., Mehta, A.D., Lakatos, P., Schroeder, C.E., 2015. Laminar profile and physiology of the alpha rhythm in primary visual, auditory, and somatosensory regions of neocortex. J. Neurosci. 35 (42), 14341–14352. Haigh, S.M., Coffman, B.A., Salisbury, D.F., 2017. Mismatch negativity in first-episode schizophrenia: a meta-analysis. Clin. EEG Neurosci. 48 (1), 3–10. Heekeren, K., Daumann, J., Neukirch, A., Stock, C., Kawohl, W., Norra, C., Waberski, T.D., Gouzoulis-Mayfrank, E., 2008. Mismatch negativity generation in the human 5HT2A agonist and NMDA antagonist model of psychosis. Psychopharmacology 199 (1), 77–88. Heresco-Levy, U., Javitt, D.C., Ebstein, R., Vass, A., Lichtenberg, P., Bar, G., Catinari, S., Ermilov, M., 2005. D-Serine efficacy as add-on pharmacotherapy to risperidone and olanzapine for treatment-refractory schizophrenia. Biol. Psychiatry 57 (6), 577–585. Hong, L.E., Moran, L.V., Du, X., O'Donnell, P., Summerfelt, A., 2012. Mismatch negativity and low frequency oscillations in schizophrenia families. Clin. Neurophysiol. 123 (10), 1980–1988. Hsiao, F.J., Wu, Z.A., Ho, L.T., Lin, Y.Y., 2009. Theta oscillation during auditory change detection: an MEG study. Biol. Psychol. 81 (1), 58–66. Insel, T.R., 2010. Rethinking schizophrenia. Nature 468 (7321), 187–193. Jahshan, C., Wynn, J.K., Green, M.F., 2013. Relationship between auditory processing and affective prosody in schizophrenia. Schizophr. Res. 143 (2–3), 348–353. Javitt, D.C., 2000. Intracortical mechanisms of mismatch negativity dysfunction in schizophrenia. Audiol. Neurootol. 5 (3–4), 207–215.

9

Javitt, D.C., 2015a. Current and emergent treatments for symptoms and neurocognitive impairment in schizophrenia. Curr. Treat. Options Psychiatry 1 (2), 107–120. Javitt, D.C., 2015b. Neurophysiological models for new treatment development in schizophrenia: early sensory approaches. Ann. N. Y. Acad. Sci. 1344, 92–104. Javitt, D.C., Freedman, R., 2015. Sensory processing dysfunction in the personal experience and neuronal machinery of schizophrenia. Am. J. Psychiatry 172 (1), 17–31. Javitt, D.C., Sweet, R.A., 2015. Auditory dysfunction in schizophrenia: integrating clinical and basic features. Nat. Rev. Neurosci. 16 (9), 535–550. Javitt, D.C., Zukin, S.R., 1991. Recent advances in the phencyclidine model of schizophrenia. Am. J. Psychiatry 148 (10), 1301–1308. Javitt, D.C., Steinschneider, M., Schroeder, C.E., Arezzo, J.C., 1996. Role of cortical N-methyl-Daspartate receptors in auditory sensory memory and mismatch negativity generation: implications for schizophrenia. Proc. Natl. Acad. Sci. U. S. A. 93 (21), 11962–11967. Javitt, D.C., Shelley, A., Ritter, W., 2000. Associated deficits in mismatch negativity generation and tone matching in schizophrenia. Clin. Neurophysiol. 111 (10), 1733–1737. Javitt, D.C., Spencer, K.M., Thaker, G.K., Winterer, G., Hajos, M., 2008a. Neurophysiological biomarkers for drug development in schizophrenia. Nat. Rev. 7 (1), 68–83. Javitt, D.C., Spencer, S., Thaker, G.K., Winterer, G., Hajos, M., 2008b. Neurophysiological biomarkers for drug development in schizophrenia. Nat. Rev. Drug Discov. 7 (1), 1–17. Javitt, D.C., Schoepp, D., Kalivas, P.W., Volkow, N.D., Zarate, C., Merchant, K., Bear, M.F., Umbricht, D., Hajos, M., Potter, W.Z., Lee, C.M., 2011. Translating glutamate: from pathophysiology to treatment. Sci. Transl. Med. 3 (102) (102mr102 PMCID: PMC3273336). Kantrowitz, J.T., Javitt, D.C., 2010a. N-methyl-D-aspartate (NMDA) receptor dysfunction or dysregulation: the final common pathway on the road to schizophrenia? Brain Res. Bull. 83 (3–4), 108–121 (PMC2941541). Kantrowitz, J.T., Javitt, D.C., 2010b. Thinking glutamatergically: changing concepts of schizophrenia based upon changing neurochemical models. Clin. Schizophr. Relat. Psychoses 4 (3), 189–200. Kantrowitz, J.T., Malhotra, A.K., Cornblatt, B., Silipo, G., Balla, A., Suckow, R.F., D'Souza, C., Saksa, J., Woods, S.W., Javitt, D.C., 2010. High dose D-serine in the treatment of schizophrenia. Schizophr. Res. 121 (1–3), 125–130. Kantrowitz, J.T., Hoptman, M.J., Leitman, D.I., Silipo, G., Javitt, D.C., 2014. The 5% difference: early sensory processing predicts sarcasm perception in schizophrenia and schizoaffective disorder. Psychol. Med. 44 (1), 25–36. Kantrowitz, J.T., Hoptman, M.J., Leitman, D.I., Moreno-Ortega, M., Lehrfeld, J.M., Dias, E., Sehatpour, P., Laukka, P., Silipo, G., Javitt, D.C., 2015a. Neural substrates of auditory emotion recognition deficits in schizophrenia. J. Neurosci. 35 (44), 14909–14921. Kantrowitz, J.T., Woods, S.W., Petkova, E., Cornblatt, B., Corcoran, C.M., Chen, H., Silipo, G., Javitt, D.C., 2015b. D-Serine for the treatment of negative symptoms in individuals at clinical high risk of schizophrenia: a pilot, double-blind, placebo-controlled, randomised parallel group mechanistic proof-of-concept trial. Lancet Psychiatry 2 (5), 403–412. Kantrowitz, J.T., Epstein, M.L., Beggel, O., Rohrig, S., Lehrfeld, J.M., Revheim, N., Lehrfeld, N.P., Reep, J., Parker, E., Silipo, G., Ahissar, M., Javitt, D.C., 2016. Neurophysiological mechanisms of cortical plasticity impairments in schizophrenia and modulation by the NMDA receptor agonist D-serine. Brain J. Neurol. 139 (Pt 12), 3281–3295. Kantrowitz, J.T., Nolan, K.A., Epstein, M., et al., 2017. Neurophysiological effects of bitopertin in schizophrenia. J. Clin. Psychopharmacol. submitted for publication. Kay, S.R., Fiszbein, A., Opler, L.A., 1987. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr. Bull. 13 (2), 261–276. Kayser, J., Tenke, C.E., Kroppmann, C.J., Alschuler, D.M., Fekri, S., Ben-David, S., Corcoran, C.M., Bruder, G.E., 2014. Auditory event-related potentials and alpha oscillations in the psychosis prodrome: neuronal generator patterns during a novelty oddball task. Int. J. Psychophysiol. 91 (2), 104–120. Ko, D., Kwon, S., Lee, G.T., Im, C.H., Kim, K.H., Jung, K.Y., 2012. Theta oscillation related to the auditory discrimination process in mismatch negativity: oddball versus control paradigm. J. Clin. Neurol. (Seoul, Korea) 8 (1), 35–42. Krystal, J.H., Karper, L.P., Seibyl, J.P., Freeman, G.K., Delaney, R., Bremner, J.D., Heninger, G.R., Bowers Jr., M.B., Charney, D.S., 1994. Subanesthetic effects of the noncompetitive NMDA antagonist, ketamine, in humans. Psychotomimetic, perceptual, cognitive, and neuroendocrine responses. Arch. Gen. Psychiatry 51 (3), 199–214. Lakatos, P., Schroeder, C.E., Leitman, D.I., Javitt, D.C., 2013. Predictive suppression of cortical excitability and its deficit in schizophrenia. J. Neurosci. 33 (28), 11692–11702. Lane, H.Y., Chang, Y.C., Liu, Y.C., Chiu, C.C., Tsai, G.E., 2005. Sarcosine or D-serine add-on treatment for acute exacerbation of schizophrenia: a randomized, double-blind, placebo-controlled study. Arch. Gen. Psychiatry 62 (11), 1196–1204. Lane, H.Y., Lin, C.H., Huang, Y.J., Liao, C.H., Chang, Y.C., Tsai, G.E., 2010. A randomized, double-blind, placebo-controlled comparison study of sarcosine (N-methylglycine) and D-serine add-on treatment for schizophrenia. Int. J. Neuropsychopharmacol. 13 (4), 451–460. Lane, H.Y., Lin, C.H., Green, M.F., et al., 2013. Add-on treatment of benzoate for schizophrenia: a randomized, double-blind, placebo-controlled trial of D-amino acid oxidase inhibitor. JAMA Psychiat. 70 (12), 1267–1275. Lavoie, S., Murray, M.M., Deppen, P., Knyazeva, M.G., Berk, M., Boulat, O., Bovet, P., Bush, A.I., Conus, P., Copolov, D., Fornari, E., Meuli, R., Solida, A., Vianin, P., Cuenod, M., Buclin, T., Do, K.Q., 2008. Glutathione precursor, N-acetyl-cysteine, improves mismatch negativity in schizophrenia patients. Neuropsychopharmacology 33 (9), 2187–2199. Lee, M., Hoptman, M.J., Lakatos, P., Kantrowitz, J.T., Dias, E., Martinez, A.M., Javitt, D.C., 2017. Neural basis of mismatch negativity (MMN) dysfunction in schizophrenia: circuit and cellular level of analysis. Mol. Psychiatry http://dx.doi.org/10.1038/mp.2017.3. Leung, S., Croft, R.J., Baldeweg, T., Nathan, P.J., 2007. Acute dopamine D(1) and D(2) receptor stimulation does not modulate mismatch negativity (MMN) in healthy human subjects. Psychopharmacology 194 (4), 443–451.

Please cite this article as: Kantrowitz, J.T., et al., Improvement in mismatch negativity generation during d-serine treatment in schizophrenia: Correlation with symptoms, Schizophr. Res. (2017), http://dx.doi.org/10.1016/j.schres.2017.02.027

10

J.T. Kantrowitz et al. / Schizophrenia Research xxx (2017) xxx–xxx

Light, G.A., Swerdlow, N.R., 2015. Future clinical uses of neurophysiological biomarkers to predict and monitor treatment response for schizophrenia. Ann. N. Y. Acad. Sci. 1344, 105–119. Light, G.A., Swerdlow, N.R., Rissling, A.J., Radant, A., Sugar, C.A., Sprock, J., Pela, M., Geyer, M.A., Braff, D.L., 2012. Characterization of neurophysiologic and neurocognitive biomarkers for use in genomic and clinical outcome studies of schizophrenia. PLoS One 7 (7), e39434. Light, G.A., Swerdlow, N.R., Thomas, M.L., et al., 2015. Validation of mismatch negativity and P3a for use in multi-site studies of schizophrenia: characterization of demographic, clinical, cognitive, and functional correlates in COGS-2. Schizophr. Res. 163 (1-3), 63–72. Lin, C.Y., Liang, S.Y., Chang, Y.C., Ting, S.Y., Kao, C.L., Wu, Y.H., Tsai, G.E., Lane, H.Y., 2015. Adjunctive sarcosine plus benzoate improved cognitive function in chronic schizophrenia patients with constant clinical symptoms: a randomised, double-blind, placebo-controlled trial. World J. Biol. Psychiatry 1–12. Lindenmayer, J.P., Bernstein-Hyman, R., Grochowski, S., 1994. A new five factor model of schizophrenia. Psychiatry Q. 65 (4), 299–322. Lisman, J.E., Coyle, J.T., Green, R.W., Javitt, D.C., Benes, F.M., Heckers, S., Grace, A.A., 2008. Circuit-based framework for understanding neurotransmitter and risk gene interactions in schizophrenia. Trends Neurosci. 31 (5), 234–242 (PMCID: PMC2680493). Luck, S.J., Mathalon, D.H., O'Donnell, B.F., Hamalainen, M.S., Spencer, K.M., Javitt, D.C., Uhlhaas, P.J., 2011. A roadmap for the development and validation of event-related potential biomarkers in schizophrenia research. Biol. Psychiatry 70 (1), 28–34. Mantysalo, S., Naatanen, R., 1987. The duration of a neuronal trace of an auditory stimulus as indicated by event-related potentials. Biol. Psychol. 24 (3), 183–195. Moghaddam, B., Krystal, J.H., 2012. Capturing the angel in “angel dust”: twenty years of translational neuroscience studies of NMDA receptor antagonists in animals and humans. Schizophr. Bull. 38 (5), 942–949. Naatanen, R., Sussman, E.S., Salisbury, D., Shafer, V.L., 2014. Mismatch negativity (MMN) as an index of cognitive dysfunction. Brain Topogr. 27 (4), 451–466. Nuechterlein, K.H., Green, M.F., 2006. MATRICS Consensus Cognitive Battery. MATRICS Assessment, Inc., Los Angeles. Oostenveld, R., Fries, P., Maris, E., Schoffelen, J.M., 2011. FieldTrip: open source software for advanced analysis of MEG, EEG, and invasive electrophysiological data. Comput. Intell. Neurosci. 2011, 156869. Potes, C., Brunner, P., Gunduz, A., Knight, R.T., Schalk, G., 2014. Spatial and temporal relationships of electrocorticographic alpha and gamma activity during auditory processing. NeuroImage 97, 188–195. Rosburg, T., Kreitschmann-Andermahr, I., 2016. The effects of ketamine on the mismatch negativity (MMN) in humans—a meta-analysis. Clin. Neurophysiol. 127 (2), 1387–1394. Rowland, L.M., Summerfelt, A., Wijtenburg, S.A., Du, X., Chiappelli, J.J., Krishna, N., West, J., Muellerklein, F., Kochunov, P., Hong, L.E., 2016. Frontal glutamate and gammaaminobutyric acid levels and their associations with mismatch negativity and digit sequencing task performance in schizophrenia. JAMA Psychiat. 73 (2), 166–174. Simpson, G.M., Angus, J.W.S., 1970. A rating scale for extrapyramidal side effects. Acta Psychiatr. Scand. 212 (Suppl), 11–19. Thomas, M.L., Green, M.F., Hellemann, G., et al., 2017. Modeling Deficits From Early Auditory Information Processing to Psychosocial Functioning in Schizophrenia. JAMA Psychiat. 74 (1), 37–46. Todd, J., Michie, P.T., Schall, U., Karayanidis, F., Yabe, H., Naatanen, R., 2008. Deviant matters: duration, frequency, and intensity deviants reveal different patterns of

mismatch negativity reduction in early and late schizophrenia. Biol. Psychiatry 63 (1), 58–64. Todd, J., Michie, P.T., Schall, U., Ward, P.B., Catts, S.V., 2012. Mismatch negativity (MMN) reduction in schizophrenia-impaired prediction—error generation, estimation or salience? Int. J. Psychophysiol. 83 (2), 222–231. Tsai, G.E., Lin, P.Y., 2010. Strategies to enhance N-methyl-D-aspartate receptor-mediated neurotransmission in schizophrenia, a critical review and meta-analysis. Curr. Pharm. Des. 16 (5), 522–537. Tsai, G., Yang, P., Chung, L.C., Lange, N., Coyle, J.T., 1998. D-Serine added to antipsychotics for the treatment of schizophrenia. Biol. Psychiatry 44 (11), 1081–1089. Tsai, G.E., Yang, P., Chung, L.C., Tsai, I.C., Tsai, C.W., Coyle, J.T., 1999. D-Serine added to clozapine for the treatment of schizophrenia. Am. J. Psychiatry 156 (11), 1822–1825. Tsai, G.E., Yang, P., Chang, Y.C., Chong, M.Y., 2006. D-Alanine added to antipsychotics for the treatment of schizophrenia. Biol. Psychiatry 59, 230–234. Umbricht, D., Krljes, S., 2005. Mismatch negativity in schizophrenia: a meta-analysis. Schizophr. Res. 76 (1), 1–23. Umbricht, D., Javitt, D., Novak, G., Bates, J., Pollack, S., Lieberman, J., Kane, J., 1998a. Effects of clozapine on auditory event-related potentials in schizophrenia. Biol. Psychiatry 44 (8), 716–725. Umbricht, D., Vollenweider, F.X., Javitt, D.C., 1998b. The effects of ketamine on mismatch negativity (MMN). Abstracts, First International Workshop on Mismatch Negativity and its Clinical Applications. 1, p. 85. Umbricht, D., Javitt, D., Novak, G., Bates, J., Pollack, S., Lieberman, J., Kane, J., 1999. Effects of risperidone on auditory event-related potentials in schizophrenia. Int. J. Neuropsychopharmacol. 2 (4), 299–304. Umbricht, D., Schmid, L., Koller, R., Vollenweider, F.X., Hell, D., Javitt, D.C., 2000. Ketamineinduced deficits in auditory and visual context-dependent processing in healthy volunteers: implications for models of cognitive deficits in schizophrenia. Arch. Gen. Psychiatry 57 (12), 1139–1147. Umbricht, D., Alberati, D., Martin-Facklam, M., et al., 2014. Effect of bitopertin, a glycine reuptake inhibitor, on negative symptoms of schizophrenia: a randomized, doubleblind, proof-of-concept study. JAMA Psychiat. 71 (6), 637–646. Viechtbauer, W., 2010. Conducting meta-analyses in R with the metafor package. J. Stat. Softw. 36 (3). Wacongne, C., 2016. A predictive coding account of MMN reduction in schizophrenia. Biol. Psychol. 116, 68–74. Wacongne, C., Changeux, J.P., Dehaene, S., 2012. A neuronal model of predictive coding accounting for the mismatch negativity. J. Neurosci. 32 (11), 3665–3678. Weiser, M., Heresco-Levy, U., Davidson, M., Javitt, D.C., Werbeloff, N., Gershon, A.A., Abramovich, Y., Amital, D., Doron, A., Konas, S., Levkovitz, Y., Liba, D., Teitelbaum, A., Mashiach, M., Zimmerman, Y., 2012. A multicenter, add-on randomized controlled trial of low-dose D-serine for negative and cognitive symptoms of schizophrenia. J. Clin. Psychiatry 73 (6), e728–e734. Womelsdorf, T., Valiante, T.A., Sahin, N.T., Miller, K.J., Tiesinga, P., 2014. Dynamic circuit motifs underlying rhythmic gain control, gating and integration. Nat. Neurosci. 17 (8), 1031–1039. Woods, S.W., 2003. Chlorpromazine equivalent doses for the newer atypical antipsychotics. J. Clin. Psychiatry 64 (6), 663–667.

Please cite this article as: Kantrowitz, J.T., et al., Improvement in mismatch negativity generation during d-serine treatment in schizophrenia: Correlation with symptoms, Schizophr. Res. (2017), http://dx.doi.org/10.1016/j.schres.2017.02.027