Psychiatry Research 169 (2009) 124–131

Contents lists available at ScienceDirect

Psychiatry Research

j ourna l homepage: www.e lsev ie r.com/ locate /psychres

Comparative effectiveness of biomarkers and clinical indicators for predictingoutcomes of SSRI treatment in Major Depressive Disorder: Results of theBRITE-MD study

Andrew F. Leuchter a,b,⁎, Ian A. Cook a,b, Lauren B. Marangell c, William S. Gilmer d, Karl S. Burgoyne b,e,Robert H. Howland f, Madhukar H. Trivedi g, Sidney Zisook h, Rakesh Jain i, James T. McCracken b,Maurizio Fava j, Dan Iosifescu j, Scott Greenwald k

a Laboratory of Brain, Behavior, and Pharmacology, Semel Institute for Neuroscience and Human Behavior at UCLA, Los Angeles, CA, United Statesb Department of Psychiatry and Biobehavioral Sciences, David Geffen School of Medicine at UCLA, Los Angeles, CA, United Statesc Department of Psychiatry, Baylor College of Medicine, Houston, TX, United Statesd Department of Psychiatry, Northwestern University, Chicago, IL, United Statese Department of Psychiatry, Harbor-UCLA Medical Center, Los Angeles, CA, United Statesf Western Psychiatric Institute and Clinic, University of Pittsburgh, Pittsburgh, PA, United Statesg Department of Psychiatry, University of Texas Southwestern Medical School, Dallas, TX, United Statesh Department of Psychiatry, UCSD, San Diego, CA, United Statesi Psychiatry, RD Clinical Research, Houston, TX, United Statesj Department of Psychiatry, Massachusetts General Hospital, Boston, MA, United Statesk Neuroscience, Aspect Medical Systems, Norwood, MA, United States

⁎ Corresponding author. Laboratory of Brain, BehavInstitute for Neuroscience and Human Behavior at UCLABiobehavioral Sciences, David Geffen School of MedicineRm. 37-452, Los Angeles, CA 90024-1759, United States.310 825 7642.

E-mail address: afl@ucla.edu (A.F. Leuchter).

0165-1781/$ – see front matter © 2009 Elsevier Irelanddoi:10.1016/j.psychres.2009.06.004

a b s t r a c t

a r t i c l e i n f o

Article history:Received 15 January 2009Received in revised form 1 June 2009Accepted 11 June 2009

Keywords:Major depressionEscitalopramPredictors of treatment responseGenetic polymorphismsQuantitative electroencephalographyAntidepressant Treatment Response (ATR)index

Patients with Major Depressive Disorder (MDD) may not respond to antidepressants for 8 weeks or longer. Abiomarker that predicted treatment effectiveness after only 1 week could be clinically useful. We examined afrontal quantitative electroencephalographic (QEEG) biomarker, the Antidepressant Treatment Response (ATR)index, as a predictor of response to escitalopram, and compared ATR with other putative predictors. Threehundred seventy-five subjects meeting DSM-IV criteria for MDD had a baseline QEEG study. After 1 week oftreatment with escitalopram, 10 mg, a second QEEG was performed, and the ATR was calculated. Subjects thenwere randomlyassigned to continuewith escitalopram,10mg, or change to alternative treatments. Seventy-threeevaluable subjects received escitalopram for a total of 49 days. Response and remission rates were 52.1% and38.4%, respectively. The ATR predicted both response and remissionwith 74% accuracy. Neither serum drug levelsnor 5HTTLPR and 5HT2a genetic polymorphisms were significant predictors. Responders had larger decreasesin Hamilton Depression Rating Scale (Ham-D17) scores at day 7 (P=0.005), but remitters did not. Clinicianprediction based upon global impression of improvement at day 7 did not predict outcome. Logistic regressionshowed that the ATR and early Ham-D17 changes were additive predictors of response, but the ATR was the onlysignificant predictor of remission. Future studies should replicate these results prior to clinical use.

© 2009 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

Major Depressive Disorder (MDD) is a leading cause of disabilitywith total costs to society in excess of $80billion annually;approximately two-thirds of these costs reflect the enormousdisability associated with MDD (Greenberg et al., 2003; Kessleret al., 2006, 2003, 1994). One reason for these high costs is the length

ior, and Pharmacology, Semel, Department of Psychiatry andat UCLA, 760 Westwood Plaza,Tel.: +1 310 825 0207; fax: +1

Ltd. All rights reserved.

of time it takes for patients to recover. Although controlled efficacytrials suggest that most patients respond to treatment within 8 weeks(Papakostas et al., 2007), the Sequenced Treatment Alternatives toRelieve Depression (STAR⁎D) trial found that fewer than 50% ofpatients responded to the first trial of a serotonin selective reuptakeinhibitor (SSRI) antidepressant (citalopram) and fewer than one-thirdachieved remission (Trivedi et al., 2006). Under standard care, theproportion of patients responding and remitting usually is even lower(Katon et al., 1996, 1999; Trivedi et al., 2004). Consequently, achievingresponse or remission with an initial medication remains a challengefor most patients with MDD and their physicians.

At present, there is no reliable method for predicting whether amedication will lead to response or remission other than “watchfulwaiting.” Methods to predict which medication would most likely

125A.F. Leuchter et al. / Psychiatry Research 169 (2009) 124–131

benefit an individual patient could reduce patients' suffering. Such toolsmight include clinical features, biomarkers such as brain-imagingfindings, or genetic polymorphisms (Bearden and Freimer, 2006).

Clinical characteristics have the advantage of being relatively easyto determine, but generally have not been useful for predictingresponse to particular medications. Symptom clusters such as anxietyor melancholia are associated with the overall likelihood of recoverybut have not been shown to be reliable predictors of response to aspecific medication for an individual patient (Fava et al., 2008; Rush,2007; Small et al., 1995; Trivedi et al., 2006). Brain imaging also hasbeen shown to have some promise for predicting response totreatment. Data suggest that pretreatment cerebral metabolism,white-matter lesions, or atrophy may be associated with outcome(Konarski et al., 2007), but the burden and cost of these procedureshave limited their clinical adoption. Some genetic biomarkers, mostnotably genetic polymorphisms in the serotonin system, have beenshown to influence the outcome of SSRI treatment. Two common andpromising candidate polymorphisms are those in the promoter regionof the serotonin transporter (5HTTLPR) and in the 5HT2a postsynapticreceptor, which in some studies have been associated with treatmentresponse (Anguelova et al., 2003; McMahon et al., 2006).

One biomarker that has promise as a predictor of treatmentresponse is quantitative electroencephalography (QEEG). QEEG powerin the theta and alpha frequency bands (Knott et al., 1996; Ulrich et al.,1994, 1988) may identify patients who are most likely to respond totricyclic antidepressants (TCAs) or SSRIs. Recent studies found thatQEEG changes in the prefrontal region may reliably identify anti-depressant medication responders within the first week of treatment(Cook et al., 2002; Leuchter et al., 1999). These findings are consistentwith the fact that rhythmic midline prefrontal EEG activity has beenshown to reflect the activity of the anterior cingulate and midlineprefrontal cortex (Asada et al., 1999), brain areas implicated in moodregulation and the pathogenesis of depression. Refinement of thismethod might permit use of a limited electrode array in the prefrontalregion (Iosifescu et al., 2006; Leuchter et al., 2005; Poland et al., 2006)that would be practical for routine clinical use.

The Biomarkers for Rapid Identification of Treatment Effectiveness inMajor Depression (BRITE-MD) study was designed to evaluate severalpossible biomarkers and clinical measures that might be useful to helpdirect antidepressant medication decisions. The protocol assessed thepredictive value of a frontal QEEG parameter, the AntidepressantTreatment Response (ATR) index (Aspect Medical Systems; Norwood,MA), which incorporates several EEG features determined frompreviously collected EEG datasets to be associated with response and/or remission during antidepressant treatment (Cook et al., 2002;Iosifescu et al., 2006; Leuchter et al., 2008). In this initial report fromBRITE-MD, we tested the primary hypothesis that the ATR at 1 weekafter initiation of treatment with the SSRI escitalopram would predictresponse and remission after 7 weeks of treatment. We further testedthe hypothesis that early changes in depressive symptom ratings,5HTTLPR and 5HT2a genetic polymorphisms, and escitalopram serumlevels, aswell as investigator predictionsbasedupon clinical impression,also would predict treatment response and remission.

2. Methods

2.1. Overview

The BRITE-MD study (ClinicalTrials.gov NCT00289523) was conducted at nine sites(departments of psychiatry at Baylor College of Medicine, Harbor-UCLAMedical Center,Massachusetts General Hospital, Northwestern University, UCLA Westwood, UCSD,University of Pittsburgh, and University of Texas Southwestern, as well as RD ClinicalResearch, a freestanding research facility). Institutional Review Boards approved themethods of the study.

2.2. Subjects

Three hundred seventy-five subjects, 18–75 years of age, who met the DSM-IVcriteria forMajor Depressive Disorder, based on theMini International Neuropsychiatric

Interview (MINI) (Sheehan et al., 1997), were enrolled in the study. All subjects had aQuick Inventory of Depressive Symptomatology-Self Rated version (QIDS-SR16) (Rushet al., 2003) score≥12, were in good physical health (i.e., free of any medical conditionsufficiently serious to affect brain function), and had no history of seizures, brainsurgery, skull fracture, significant head trauma, or previous abnormal EEG. All subjectsgave informed consent prior to assessment or any study procedures.

Subjects were excluded from the study if they could not give informed consent, werepregnant or refused to use medically acceptable birth control during the study, metcriteria for bipolar or psychotic disorder or substance dependence or abuse within thepast 6 months, suffered from cognitive disorder, or met criteria for Axis II cluster A or Bdiagnosis sufficiently severe to interfere with completion of the protocol. Subjects alsowere excluded if they had failed to benefit from an adequate trial of treatment or failed totolerate either of the study medications during the current episode, had a course of ECTwithin the past 6 months, had a contraindication for use of either of the study drugs, hadbeen treated with fluoxetine or a monoamine oxidase inhibitor (MAOI) within the past 4weeks, were clinically stable on current antidepressant medication(s) or had startedspecific psychotherapy for depression (i.e., CBT, IPT) within the past 2 months. Subjectswere tested and excluded for use of illicit substances or certain other central nervoussystem active medications within 1 week prior to enrollment, including antidepressants,anticonvulsants/mood stabilizers, anticholinergics, antipsychotics, migraine medications,Parkinsonism medications, barbiturates, benzodiazepines, herbal preparations, musclerelaxants, psychostimulants, and systemic corticosteroids. Medications acceptable foroccasional use (not within 48h of a QEEG) included non-sedating antihistamines,codeine- or oxycodone-containing compounds, over-the-counter cold remedies, coughsuppressants, and non-prescription sleep aids. After complete description of the study tothe subjects, written informed consent was obtained.

2.3. Treatment

Study medications were administered in an open-label manner. Subjects receivedescitalopram, 10 mg daily, for 1 week, after which time they were randomized either tocontinue escitalopram, 10 mg (ESC; primary study arm), switch to bupropion XL,300 mg (BUP), or combine escitalopram, 10 mg, with bupropion XL, 300 mg (COMB).For this initial report, we present only the results for the ESC group through the primaryendpoint because this was the group that received continuous treatment with a singleagent throughout the trial, and for which the clinical symptom changes in the first weekof treatment would be most interpretable. Treatment continued at this dosage through7 weeks (day 49) (1 week of initial treatment with escitalopram plus 6 weeks afterrandomization), the primary study endpoint (Fig. 1). If reduction in dose was clinicallyindicated, the subject was removed from the study.

If the subject achieved remission at the primary endpoint, the escitalopram wascontinued at the same dosage, but if the 17-item Hamilton Depression Rating Scale(Ham-D17) score remained N7, escitalopram could be increased to 20 mg qd no laterthan day 53 and dosage continued as tolerated through 91 days of treatment (total13 weeks). At the end of week 13, the Ham-D17 and IDS scores were assessed.

2.4. Assessment

All subjects underwent diagnostic assessment with the MINI. Subjects over 60 alsowere assessed with the Mini Mental State Examination (MMSE) (Folstein et al., 1975)and those with an MMSE≤24 were evaluated by a study physician using a DSM-IVchecklist for dementia. Eligibility for the study alsowas determined using the QIDS-SR16

as described above.The primary efficacy measure was the Ham-D17 (Hamilton, 1960) assessed at 7

weeks (day 49). Response was defined as a decrease in the Ham-D17≥50% from thebaseline value and remission as a Ham-D17≤7. Severity of depression at baseline alsowas assessed using the Inventory of Depressive Symptomatology-Clinician-rated (IDS-C30) (Rush et al., 1996) and the Ham-D17 to measure core diagnostic and commonlyassociated symptoms of depression. The IDS-C and Ham-D17 were administered using acombined structured interview guide (www.ids-qids.org). Severity of illness also wasassessed using the Clinical Global Impression Scale (CGI) (Guy, 1976), and at day 7 astudy physician used a modified CGI to make a clinical prediction of the likely degree ofbenefit that each subject would obtain from 6 weeks of escitalopram treatment: 0=nosignificant predicted benefit, 1=predicted improvement but not response, 2=pre-dicted response but not remission, and 3=predicted remission.

2.5. EEG biomarker methods

EEG data were collected using Aspect Medical Systems' NS-5000 system. Thisconsisted of a PC-compatible laptop computer connected to a four-channel EEGacquisition device (BIS×4) that performed digitization as well as signal filtering andconditioning, connected through a shielded cable to six self-prepping electrodes(Zipprep™) (Aspect Medical Systems; Norwood, MA) applied at four recording sites onthe forehead (Fpz, FT7, FT8, ground) and two on the earlobes (A1, A2) (Fig. 2). EEG datawere recorded while the subject rested in a reclining chair during two 6-min eyes-closed segments, separated by a 2-min eyes-open segment.

Following rejection of artifact, power spectra of the EEG (A1-Fpz, A2-Fpz) werecalculated using 2-s epochs of an eyes-closed resting period. Values were calculatedseparately for each channel in each epoch and then averaged for the two channels. ATRis a non-linear weighted combination of three EEG features, measured at baseline and 1

Fig. 1. Study flow chart.

126 A.F. Leuchter et al. / Psychiatry Research 169 (2009) 124–131

week after the start of treatment, that previously were identified as being associatedwith antidepressant outcome (Cook et al., 2002; Iosifescu et al., 2006; Poland et al.,2006). These three features are relative combined theta and alpha power (3–12 Hz),alpha1 absolute power (8.5–12 Hz), and alpha2 absolute power (9–11.5 Hz). Relativecombined theta and alpha power (3–12 Hz) is calculated as the ratio of absolutecombined theta and alpha power divided by total power (2–20 Hz). The ATR is aweighted combination of the relative theta and alpha power at week 1, and thedifference between alpha1 power at baseline and alpha2 power at week 1, scaled torange from 0 (low probability) to 100 (high probability of response).

The ATR was evaluated in the BRITE-MD through a split-half data analysis designspecified a priori in the protocol. The initial weighted combination of the features in theATR (version 4.0) was determined based upon previous datasets collected with multipledifferent recording systems and electrodeplacements. TheBRITEprotocol specified that aninterim analysis would be performed at the study midpoint to assess the accuracy of the

Fig. 2. Sites of electrode placement.

ATR 4.0 in predicting treatment response with the new electrode placements andrecording systemused in this study. Based upon the interimanalysis, the combinationwasadjusted (version 4.1) to maximize accuracy on the first half of the dataset. ATR 4.1 thenwas tested independently on the second half of the dataset without further modification,and subsequently on the entire dataset. The results reported here were calculated usingATR 4.1 on the entire dataset. Further details regarding the calculations of the ATR,including results of the split-half analyses, are reported separately in Appendix A.

2.6. Serum drug levels

Escitalopram and desmethylescitalopram plasma levels were determined usinghigh performance liquid chromatography. Assays were performed in a commerciallaboratory (Pacific Toxicology Laboratories, Inc.; Chatsworth, CA) using samples drawnafter 1 week of treatment with escitalopram.

2.7. Genotyping

DNA was extracted from whole blood following Gentra DNA collection procedures.The sequence of DNA at the HTR2A rs7997012 (G/A) single nucleotide polymorphism(SNP)was amplified using primer sequences TCTAATCTAACTTCTGCATACTCAGAACAG andCTCAGAGGATGTTCTCCTTGGAGGCACAGC designed from the sequence listed by theNational Center of Biotechnology Information (NCBI) (rs7997012.22) (McMahon et al.,2006). The PCR assay was performed using Qiagen HotStarTaq solution kit (Qiagen USA;Valencia, CA) and a PCR programof 95 °C for 15 min, 35 cycles of 95 °C (30 s), 59 °C (30 s),and 72 °C (30 s), followed by 72 °C (10 min). The product (383 bps) was digested by 10Uof PacI (New England Biolabs; Ipswich,MA)with corresponding buffer at 37 °C for 3h. Theenzyme selectively digested the A allele, yielding two fragments of 104 and 179 bps. Gelsincluding relevant molecular weight standards and controls of known genotype werescored by two readers. Genotypes were assessed for Hardy–Weinberg (HW) equilibriumwith the goodness-of-fit chi-square test using SAS PROC ALLELE.

5HTTLPR primer sequences (forward: 5′-GGC GTT GCC GCT CTG AAT GC-3′; reverse:5″-GAGGGACTGAGC TGGACAACCAC-3′) were used to generate a 484/528-bp fragment.PCRwasperformed inamixture containing10×buffer, Q solution, 200 μmol/l of eachdATP,dCTP, dTTP, andGTP,1 μmol/l of eachprimer, 0.6 units ofQiagenHotStarTaq, 50 nggenomicDNA, and buffer. Amplification occurred using a MCJ Tetrad with 40cycles of 30 s at 95°C,30 s at 61°C, and 1 min at 72°C, followedby 10 min at 72 °C; 10 μl of productwas separatedon2%agarose gels using knownmarker ladders (Kimet al., 2002). Gelswere scoredby tworeaders. Alleles were determined to be in HW equilibrium.

Table 1Baseline characteristics.

Metric Study arm All subjects

ESC BUP COMB(n=73) (n=73) (n=74) (n=220)

HAMD-17 20.6±4.4 21.7±4.0 20.4±4.3 20.9±4.3Age (years) 42.7±12.7 41.6±13.4 44.4±13.6 42.9±13.2Height (in) 66.5±4.3 66.7±4.1 66.8±4.5 66.7±4.3Weight (lb) 183.9±48.0 182.1±54.9 180.7±40.9 182.2±48.1Heart rate (bpm) 69.5±9.7 72.7±10.7 70.8±11.2 71.0±10.6Systolic blood pressure(mm Hg)

122.1±14.4 122.7±15.3 121.6±18.6 122.1±16.1

Diastolic blood pressure(mm Hg)

77.8±7.9 78.1±8.9 77.3±10.5 77.7±9.1

Gender (% female) 65.8 61.6 59.5 62.3Race (%)White 61.6 61.6 64.9 62.7Hispanic/Latino 21.9 21.9 14.9 19.5Black/African American 12.3 13.7 17.6 14.5Asian 4.1 1.4 2.7 2.7Other 0.0 1.4 0.0 0.5

127A.F. Leuchter et al. / Psychiatry Research 169 (2009) 124–131

2.8. Data analysis

Statistical analyses were performed using SPSS (SPSS, Inc.; Chicago, IL). Mean ATRvalues were compared between outcome groups using Student's T-test. ReceiverOperating Characteristic (ROC) analysis was used to model the sensitivity vs. (1−specificity) for the ATR as the discrimination threshold for predicting response toescitalopram was varied. A threshold was chosen on the ROC curve that optimized theoverall accuracy of the ATR in predicting response versus non-response. The primaryhypothesis of the study was that subjects with ATR values above the threshold (i.e.,“positive” biomarker) would have a significantly higher rate of clinical response andremission at the end of 7 weeks of ESC treatment than those with ATR values below thethreshold (i.e., “negative” biomarker). This hypothesis was tested using a chi-squarestatistic with ATR biomarker category as the independent variable and responsecategory as the dependent variable. A separate analysis was performed with remissioncategory as the dependent variable.

We also compared the predictive accuracy of the ATR with changes in Ham-D17

scores and clinician prediction of the likelihood of medication effectiveness at 1 week.Mean percentage change in Ham-D17 scores at 1 week were compared betweenoutcome groups using Student's T-test. Accuracy of clinician prediction of outcomegroups was evaluated using Fisher's Exact Test.

In addition, we compared the predictive accuracy of the ATR with the predictiveaccuracy of serum drug levels at 1 week, and 5HTTLPR and 5HT2a polymorphisms. Wehypothesized that the ATR would predict treatment response independently of bothserum drug levels and genetic polymorphisms. Analyses of variance (ANOVAs) wereperformed to assess whether response rate was significantly different among subjectswith different polymorphisms and whether the ATR predicted response independentlyof polymorphisms. Mean serum drug concentrations were compared between outcomegroups using Student's T-test.

Candidate predictors were considered both separately and as components of alogistic regression model (backwards conditional selection; Pin=0.05, Pout=0.10) thatexamined the relative contributions that each variable added to accuracy of prediction.Pb0.05 was considered statistically significant for all statistical tests.

3. Results

3.1. Subject characteristics

A total of 375 subjects consented to participate in the study, and331 met the criteria to participate and entered treatment. Forty-nine

Table 2Clinical and biomarker characteristics of ESC subjects by outcome group.

Mean age M:Fratio

Mean initialHam-D

Mean finalHam-D

Mean Ham-Dat day 7

%Hb

All subjects (n=73) 42.7±2.7 25:48 20.6±4.4 10.4±6.2 14.4±6.0 2Responder (n=38) 42.9±11.6 17:21 20.7±4.6 5.8±3.8 12.6±5.4 3Non-responder (n=35) 42.4±14.0 8:27 20.4±4.2 15.4±4.2 16.3±6.1 2Remitter (n=28) 44.1±10.5 13:15 19.4±3.9 3.9±1.9 12.2±5.2 3Non-remitter (n=45) 41.8±13.9 12:33 21.3±4.62 14.5±4.2 15.7±6.1 2

a Responders different from non-responders, P=0.005.b Responders different from non-responders, Pb0.001.c Remitters different from non-remitters, P=0.002.

of these subjects did not continue through the primary endpoint forreasons including changes in life circumstances and removal by theinvestigators for failure to follow the study protocol (i.e., refusal totake medication or complete rating scales). There was no difference inage, gender, or severity of baseline depression between those whocompleted treatment and thosewho discontinued prematurely. Of the282 subjects who completed through the primary endpoint, 62 wereexcluded from the analysis for ingesting excluded medications (23subjects), noncompliance (i.e., taking less than 70% of their medica-tion) (4 subjects), not having EEG recordings at specified time points(6 subjects), and EEGs containing excessive electrocardiographicartifact (29 subjects). The ATR algorithm automatically detected andexcluded subjects with this artifact in the EEG (~15% of subjects in thissample). The 220 evaluable subjects were divided relatively equallyamong the three treatment groups (73 ESC, 73 BUP, 74 COMB). Thedemographic, baseline symptom severity, and physiologic character-istics of the subjects did not differ significantly among treatmentgroups, nor between the subjects who could (n=313) and could not(n=62) be evaluated (Table 1).

3.2. Outcome of ESC treatment

The response and remission rates of the 73 ESC subjects who couldbe evaluated were 52.1% and 38.4%, respectively. Outcome groups didnot differ in their initial Ham-D17 scores or demographic factors(Table 2). There was no significant difference in response or remissionrates as a function of clinical site (data not presented).

3.3. Association between potential clinical predictors and treatmentoutcome

Responders at day 49 showed a larger mean percentage decreasein Ham-D17 scores at day 7 compared with non-responders (38±26%vs. 21±23%, P=0.005). Subjects who remitted, however, did notshow a significantly greater mean percentage reduction in Ham-D17

scores than those subjects who did not remit (36±28 vs. 26±24,P=0.12) (Table 2). Clinician prediction at day 7 of the likelihood ofresponse (51%) or remission (59%) was not significantly associatedwith outcome at the primary endpoint (PN0.44).

3.4. Association between biomarkers and treatment outcome

Responders and remitters had significantly higher ATR values thanthose who did not (Table 2). ROC analysis (Crowell et al., 2006)yielded area under the curve=0.77, Pb0.001 (Fig. 3). Based upon theROC analysis, a threshold value (58.6) was selected to maximizeaccuracy in classification of responders and non-responders totreatment: values above this threshold were designated as a positivebiomarker, and below designated as a negative biomarker. The ATRpredicted response with 74% overall accuracy, 58% sensitivity, 91%specificity, 88% positive predictive accuracy, and 67% negativepredictive accuracy. The ATR also predicted remission with 74%

Change inam-D fromaseline to day 7

Mean serumdrug level

Mean ATRvalue

Ratio (%) ofAA:AG:GG 5HT2apolymorphism

Ratio (%) ofSS:SL:LL 5HTTPLPRpolymorphism

9.7±25.9 18.4±8.1 54.6±10.2 13:46:41 23:38:397.9±25.9a 17.6±7.8 59.0±10.2b 9:41:50 31:31:370.9±23.2 19.5±8.6 49.8±7.8 18:50:32 15:44:415.7±27.8 18.0±8.5 59.2±10.7c 12:44:44 27:38:356.0±24.3 18.7±7.9 51.7±8.8 14:46:40 21:37:42

Fig. 3. Receiver Operating Characteristic curve for ATR prediction of response toescitalopram treatment.

128 A.F. Leuchter et al. / Psychiatry Research 169 (2009) 124–131

overall accuracy, 61% sensitivity, 82% specificity, 68% positivepredictive accuracy, and 77% negative predictive accuracy.

Therewasnosignificantdifference inplasma levels of escitalopram+desmethylescitalopram between those who responded or remittedwith escitalopram treatment and those who did not. ANOVA alsoshowed that there was no significant difference in the proportion ofsubjects showing different 5HTTLPR and 5HT2a polymorphismsamong those who responded or remitted (Table 2).

3.5. Combined model of predictive accuracy of biomarkers and clinicalfeatures

Predictive accuracy of change in Ham-D17 scores, clinicianprediction, 5HTTLPR or 5HT2a polymorphisms, and ATR werecompared using stepwise logistic regression. This analysis identifiedboth ATR (P=0.001) and change in Ham-D17 at 1 week (P=0.034) assignificant predictors of response. Logistic regression identified onlyATR (P=0.002) as a predictor of remission. Neither clinicianprediction nor any genetic polymorphisms entered the model forpredicting either response or remission.

4. Discussion

These results are consistent with the hypothesis that change infrontal QEEG after 1 week of treatment with a representative SSRI(escitalopram) is a useful biomarker for predicting 8-week treatmentoutcome. High ATR values were significantly associated with lowerfinal depression scores, and ATR had 74% overall accuracy in predictingresponse and remission with escitalopram treatment.

The ATR was unique among the biomarkers examined here in thatneither serum drug levels nor common genetic polymorphisms in theserotonergic system were significantly associated with response toescitalopram. It is not surprising that serum drug levels were notassociated with response because SSRI serum levels have notconsistently been associated with response (Rasmussen and Brosen,2000). While 5HTTLPR polymorphisms have been significant pre-dictors of response in several studies (Anguelova et al., 2003; Cusinet al., 2002; Peters et al., 2004), and 5HT2a polymorphisms weresignificant predictors of response in the STAR⁎D sample (McMahonet al., 2006), there has been a lack of reproducibility in manypharmacogenetic studies, probably due both to the fact that large

samples may be necessary to detect differences and that contributionsof individual genes may be small (Malhotra et al., 2004).

Decreases in severity of depressive symptoms in the first week oftreatment also predicted response to escitalopram in this study, consistentwith previous reports that early symptom changes do occur and augurwell for longer-term improvement (Posternak and Zimmerman, 2007).However, the current results indicate thatearly symptomchangesmaynotbe sufficient to direct early antidepressant treatment decisions becausethey are not closely associated with eventual remission. The currentresults also indicate that structured rating scales may be needed to detectthese changes because expert clinicians, using their global clinicalimpression, were unable to predict either response or remission beyondchance levels. Importantly, logistic regression indicated that early changesin symptoms were complementary to the ATR biomarker, and only theATR biomarker was useful for predicting remission.

The results of this study should be interpretedwithin the context offour primary limitations. First, subjects were treated with a fixed doseof escitalopram, and those receiving other psychotropic medicationswithin 1 week of the EEG, with active substance abuse, or severephysical illness or major psychiatric comorbidity were excluded. Thefindings therefore may not be generalizable to less controlledtreatment conditions or entirely naturalistic samples. Second, outcomedata from some subjects could not be used because of excessive EEGartifact, and although these subjects could not be distinguished fromthe overall subject pool on the basis of demographics or symptoms, it ispossible that exclusion of data from these subjects affected the results.Third, this study was not blinded to medication and did not include aplacebo control group. Those subjects who responded or remittedduring treatment therefore are likely to include some subjects whoimproved because of the treatment milieu as well as because of apharmacologic response. ATR measurements therefore may be inter-preted as representing changes in brain function during treatment thatincludes medication, but not necessarily entirely a medication effect.Fourth, our primary endpoint was symptom improvement at day 49 oftreatment.We cannot drawdefinite conclusions about the relationshipbetween ATR and longer-term treatment outcome.

The findings of this study are encouraging in several respects. First,they demonstrate that frontal QEEG monitoring using the ATR indexpredicts whether a representative SSRI (escitalopram) was likely to beeffective. This could be of significant benefit to patients awaiting relieffrom depressive symptoms. Information about the likelihood that amedication would be helpful could be used as one factor in makingdecisions about whether to continue or switchmedications. In addition,information indicating that a medication was likely to prove effectiveeventually might help encourage adherence during the critical firstweeks of treatment, when the risk of premature discontinuation isgreatest. Second, the study demonstrates that a greatly simplified QEEGsystem can be used to collect clinically useful data. The small number ofelectrodes utilized all were located outside the hairline, so thatrecordings were not technically demanding and could be completed in10–15 min. Third, the results indicate that the ATR index provided datathat were complementary to clinical information and, if utilizedprospectively, might enhance accuracy in early clinical decisionmaking.Further research is needed to replicate these findings, and determine ifthe ATR can help direct decision making using antidepressant medica-tions with other mechanisms of action. If the findings of this study arereplicated, frontal QEEG assessment could be considered for integrationinto clinical management of MDD.

Acknowledgments

Aspect Medical Systems provided financial support of this project. Aspectparticipated in the design and conduct of the study; collection, management, analysis,and interpretation of the data; and preparation and review of the manuscript. Finalapproval of the form and content of the manuscript rests with the authors.

The authors gratefully acknowledge the clinicians and research coordinators atBaylor College of Medicine, Harbor-UCLA Medical Center, Massachusetts GeneralHospital, Northwestern University, RD Clinical Research, the Semel Institute for

129A.F. Leuchter et al. / Psychiatry Research 169 (2009) 124–131

Neuroscience and Human Behavior at UCLA, UCSD, Western Psychiatric Institute andClinics at the University of Pittsburgh, and University of Texas Southwestern MedicalSchool for dedication to the completion of this project. We also are grateful to theNeuroscience Clinical Research Team at Aspect Medical Systems for their dedication toand support of this project. A list of other support for each investigator is provided onthe last two pages of this article.

Appendix A

The ATR was calculated from the power spectrum analysis of theEEGs performed at baseline and 1 week following the start ofantidepressant treatment, following rejection of muscle, electrocardio-graphic, and drowsiness artifact. The power spectra (A1-Fpz, A2-Fpz)were calculated using consecutive 2-s epochs of an eyes-closed restingperiod, by averaging the values calculated separately for each epoch ineach channel and then averaging values for the two channels. The ATR isa non-linear weighted combination of relative combined theta andalpha power (3–12 Hz), alpha1 absolute power (8.5–12 Hz), and alpha2absolute power (9–11.5 Hz). These three EEG features previously wereidentified as being associated with antidepressant outcome (Cook et al.,2002; Iosifescu et al., 2006; Poland et al., 2006). Relative combined thetaand alpha power (3–12 Hz) is calculated as the ratio of absolutecombined theta and alpha power divided by total power (2–20 Hz).

ATR values are calculated according to the formula:

ATR = maxf0;minð100;fA⁎ðAPðt1;αbÞ � APðt1;αaÞÞ + B⁎ðRPðt1; θ + αÞÞ + CgÞg

where ATR is the Antidepressant Treatment Response Index value;max{x,y} and min{x,y} represent the maximum or minimumvalue, respectively,of variables x and y; AP(t1,αb) is the absolute power at week one in analpha range of 9 - 11.5Hz; AP(t0,αa) is the absolute power at baseline in analpha rangeof 8.5–12Hz;RP(t1,θ+α) is the relativepoweratweekone in acombined alpha and theta band (i.e., (3–12 Hz)/(2–20 Hz)), and A, B, andC are numerical constants. ATR is bounded by values of 0 and 100.

The initialweighted combination of the features in ATR (version 4.0)wasdeterminedbaseduponpreviousdatasets thatdidnot alwaysutilizethe samemontage of recordingelectrodes. In addition, previousdatasetswere collected with multiple different recording systems. In order tostandardize ATR on the current reduced electrode placements andrecording system, the BRITE protocol specified a priori that an interimanalysis would be performed at the study midpoint to assess ATR 4.0accuracy in predicting treatment response. The interim analysis wasperformed using a split-half approach, with the results of the interimanalysis used to adjust the combination of features used to calculate ATR4.0. The analysis was performed on the first 35 subjects enrolled in theBRITE-MD study who were randomized to escitalopram treatment.Based upon the interim analysis, the combination of EEG features wasadjusted (version 4.1) to maximize accuracy on the first half of thedataset. ATR 4.1 then was prospectively evaluated on the 38 subjectswho completed escitalopram treatment following the interim analysis.The performance metrics of ATR 4.1 measured at 1 week to predict 7-week response to escitalopram treatment are tabulated below on theinterim, post-interim, and combined BRITE datasets. In addition,Receiver Operating Characteristic (ROC) curves from both halves ofthe dataset are provided. Subjects whose ATR was≥ a threshold (58.6)were predicted to respond, while those below the threshold werepredicted not to respond. ATR was a statistically significant predictor ofresponse to escitalopram in both halves of the dataset, and therewas nosignificant difference in the statistical performance metrics.

Dataset Sensitivity(%)

Specificity(%)

Positivaccura

BRITE interim analysis (n=35) 59 92 93BRITE post-interim analysis (n=38) 56 91 82All BRITE subjects (n=73) 58 91 88

Fig. A1. Interim analysis.

Fig. A2. Post-interim analysis.

To facilitate independent replication of the work reported here,Aspect intends to make available a limited number of investigationalsystems for academic researchers. Please contact ScottGreenwald, Ph.D.,at sgreenwald@aspectms.com for further information.

e predictivecy (%)

Negative predictiveaccuracy (%)

Classificationaccuracy (%)

Area underROC curve (%)

57 71 8574 76 6967 74 77

130 A.F. Leuchter et al. / Psychiatry Research 169 (2009) 124–131

References

Anguelova, M., Benkelfat, C., Turecki, G., 2003. A systematic review of associationstudies investigating genes coding for serotonin receptors and the serotonintransporter: II. Suicidal behavior. Molecular Psychiatry 8 (7), 646–653.

Asada, H., Fukuda, Y., Tsunoda, S., Yamaguchi, M., Tonoike, M., 1999. Frontal midlinetheta rhythms reflect alternative activation of prefrontal cortex and anteriorcingulate cortex in humans. Neuroscience Letters 274, 29–32.

Bearden, C.E., Freimer, N.B., 2006. Endophenotypes for psychiatric disorders: ready forprimetime? Trends in Genetics 22 (6), 306–313.

Cook, I.A., Leuchter, A.F., Morgan, M., Witte, E., Stubbeman, W.F., Abrams, M., Rosenberg,S., Uijtdehaage, S.H.J., 2002. Early changes in prefrontal activity characterize clinicalresponders to antidepressants. Neuropsychopharmacology 27 (1), 120–131.

Crowell, S.E., Beauchaine, T.P., Gatzke-Kopp, L., Sylvers, P., Mead, H., Chipman-Chacon, J.,2006. Autonomic correlates of attention-deficit/hyperactivity disorder and opposi-tional defiant disorder in preschool children. Journal of Abnormal Psychology 115(1), 174–178.

Cusin, C., Serretti, A., Zanardi, R., Lattuada, E., Rossini, D., Lilli, R., Lorenzi, C., Smeraldi, E.,2002. Influence of monoamine oxidase A and serotonin receptor 2A polymorph-isms in SSRI antidepressant activity. Int J Neuropsychopharmacol 5 (1), 27–35.

Fava, M., Rush, A.J., Alpert, J.E., Balasubramani, G.K., Wisniewski, S.R., Carmin, C.N., Biggs,M.M., Zisook, S., Leuchter, A., Howland, R., Warden, D., Trivedi, M.H., 2008.Difference in treatment outcome in outpatients with anxious versus nonanxiousdepression: a STAR⁎D report. American Journal of Psychiatry 165 (3), 342–351.

Folstein, M.F., Folstein, S.E., McHugh, P.R., 1975. qMini-mental stateq. A practical methodfor grading the cognitive state of patients for the clinician. J Psychiatry Research 12(3), 189–198.

Greenberg, P.E., Kessler, R.C., Birnbaum, H.G., Leong, S.A., Lowe, S.W., Berglund, P.A., Corey-Lisle, P.K., 2003. The economic burden of depression in the United States: how did itchange between 1990 and 2000? Journal of Clinical Psychiatry 64 (12), 1465–1475.

Guy, W., 1976. ECDEU Assessment Manual for Psychopharmacology. U.S. GovernmentPrinting Office, U.S. Dept. of Health, Education and Welfare Publication No. 76-338.Superintendent of Documents, Government Printing Office, Washington, DC.

Hamilton, M., 1960. A rating scale for depression. Journal of Neurology, Neurosurgery &Psychiatry 23, 56–62.

Iosifescu, D.V., Greenwald, S.D., Smith, C.P., Devlin, P., Alpert, J.E., Hamill, S., Fava, M.,2006. Frontal EEG at week 1 predicts clinical response to SSRI treatment in MDD.Presented as poster #231 at the Annual Meeting of the American PsychiatricAssociation. Toronto, Canada, May 20-25, 2006.

Katon, W., Robinson, P., Von Korff, M., Lin, E., Bush, T., Ludman, E., Simon, G., Walker, E.,1996. A multifaceted intervention to improve treatment of depression in primarycare. Archives of General Psychiatry 53 (10), 924–932.

Katon, W., Von Korff, M., Lin, E., Simon, G., Walker, E., Unutzer, J., Bush, T., Russo, J.,Ludman, E., 1999. Stepped collaborative care for primary care patients withpersistent symptoms of depression: a randomized trial. Archives of GeneralPsychiatry 56 (12), 1109–1115.

Kessler, R.C., Akiskal, H.S., Ames, M., Birnbaum, H., Greenberg, P., Hirschfeld, R.M., Jin, R.,Merikangas, K.R., Simon, G.E., Wang, P.S., 2006. Prevalence and effects of mooddisorders on work performance in a nationally representative sample of U.S.workers. American Journal of Psychiatry 163 (9), 1561–1568.

Kessler, R.C., Berglund, P., Demler, O., Jin, R., Koretz, D., Merikangas, K.R., Rush, A.J., Walters,E.E., Wang, P.S., National Comorbity Survey Replication, 2003. The epidemiology ofmajor depressive disorder: results from the National Comorbidity Survey Replication(NCS-R). Journal of the American Medical Association 289 (23), 3095–3105.

Kessler, R.C., McGonagle, K.A., Zhao, S., Nelson, C.B., Hughes, M., Eshleman, S., Wittchen,H.U., Kendler, K.S., 1994. Lifetime and 12-month prevalence of DSM-III-R psychiatricdisorders in the United States. Results from the National Comorbidity Survey.Archives of General Psychiatry 51 (1), 8–19.

Kim, S.J., Cox, N., Courchesne, R., Lord, C., Corsello, C., Akshoomoff, N., Guter, S.,Leventhal, B.L., Courchesne, E., Cook, E.H., Jr., 2002. Transmission disequilibriummapping at the serotonin transporter gene (SLC6A4) region in autistic disorder.Molecular Psychiatry 7 (3), 278–288.

Knott, J., Telner, J.I., Lapierre, Y.D., Browne, M., Horn, E.R., 1996. Quantitative EEG in theprediction of antidepressant response to imipramine. Journal of Affective Disorders39 (3), 175–184.

Konarski, J.Z., McIntyre, R.S., Soczynska, J.K., Kennedy, S.H., 2007. Neuroimagingapproaches in mood disorders: technique and clinical implications. Annals ofClinical Psychiatry 19 (4), 265–277.

Leuchter, A.F., Kofol, T., Greenwald, S.D., Devlin, P., Cook, I.A., 2005. Source localization ofanterior cingulate activity: comparison of 4 vs. 24 channel EEG techniques.Presented as poster at the Annual Meeting of the American Psychiatric Association.Atlanta, GA, USA, May 21-26, 2005.

Leuchter, A.F., Marangell, L.B., Burgoyne, K.S., Sigl, J.C., Jain, R., Zisook, S., McCracken, J.T.,Iosifescu, D.V., 2008. Biomarkers for Rapid Identification of Treatment Effectiveness(BRITE): testing symptoms and biomarkers as predictors of response and remissionin MDD. Presented as poster at the Annual Meeting of the American PsychiatricAssociation. Washington, D.C., USA. May 3-8.

Leuchter, A.F., Uijtdehaage, S.H., Cook, I.A., O'Hara, R., Mandelkern, M., 1999. Relation-ship between brain electrical activity and cortical perfusion in normal subjects.Psychiatry Research: Neuroimaging 90 (2), 125–140.

Malhotra, A.K., Murphy, G.M., Kennedy, J.L., 2004. Pharmacogenetics of psychotropicdrug response. American Journal of Psychiatry 161 (5), 780–796.

McMahon, F.J., Buervenich, S., Charney, D., Lipsky, R., Rush, A.J., Wilson, A.F., Sorant, A.J.,Papanicolaou, G.J., Laje, G., Fava, M., Trivedi, M.H., Wisniewski, S.R., Manji, H., 2006.Variation in the gene encoding the serotonin 2A receptor is associated with outcomeof antidepressant treatment. American Journal of Human Genetics 78 (5), 804–814.

Papakostas, G.I., Thase, M.E., Fava, M., Nelson, J.C., Shelton, R.C., 2007. Are antidepressantdrugs that combine serotonergic and noradrenergic mechanisms of action moreeffective than the selective serotonin reuptake inhibitors in treating majordepressive disorder? A meta-analysis of studies of newer agents. BiologicalPsychiatry 62 (11), 1217–1227.

Peters, E.J., Slager, S.L., McGrath, P.J., Knowles, J.A., Hamilton, S.P., 2004. Investigation ofserotonin-related genes in antidepressant response. Molecular Psychiatry 9 (9),879–889.

Poland, R., Greenwald, S.D., Smith, C.P., Kustak, C., Schulz, J., Rowe, S., Gertstik, L., 2006.Frontal EEG at 1 week predicts response to treatment with citalopram in MDD.Presented as poster at the Annual Meeting of the American Psychiatric Association.Toronto, Canada, May 20-25, 2006.

Posternak, M.A., Zimmerman, M., 2007. Therapeutic effect of follow-up assessments onantidepressant and placebo response rates in antidepressant efficacy trials: meta-analysis. British Journal of Psychiatry 190, 287–292.

Rasmussen, B.B., Brosen, K., 2000. Is therapeutic drug monitoring a case for optimizingclinical outcome and avoiding interactions of the selective serotonin reuptakeinhibitors? Therapeutic Drug Monitoring 22 (2), 143–154.

Rush, A.J., 2007. The varied clinical presentations of major depressive disorder. Journalof Clinical Psychiatry 68 (Suppl 8), 4–10.

Rush, A.J., Gullion, C.M., Basco, M.R., Jarrett, R.B., Trivedi, M.H., 1996. The Inventory ofDepressive Symptomatology (IDS): psychometric properties. Psychological Med-icine 26 (3), 477–486.

Rush, A.J., Trivedi, M.H., Ibrahim, H.M., Carmody, T.J., Arnow, B., Klein, D.N., Markowitz, J.C.,Ninan, P.T., Kornstein, S., Manber, R., Thase,M.E., Kocsis, J.H., Keller, M.B., 2003. The 16-Item Quick Inventory of Depressive Symptomatology (QIDS), clinician rating (QIDS-C), and self-report (QIDS-SR): a psychometric evaluation in patients with chronicmajor depression. Biological Psychiatry 54 (5), 573–583.

Sheehan, D.V., Lecrubier, Y., Harnett-Sheehan, K., Janavs, J., Weiller, E., Bonara, L.I.,Keskiner, A., Schinka, J., Knapp, E., Sheehan, M.F., Dunbar, G.C., 1997. Reliability andvalidity of the MINI International Neuropsychiatric Interview (M.I.N.I.): accordingto the SCID-P. European Psychiatry 12, 232–241.

Small, G.W., Hamilton, S.H., Bystritsky, A., Meyers, B.S., Nemeroff, C.B., 1995. Clinicalresponse predictors in a double-blind, placebo-controlled trial of fluoxetine forgeriatric major depression. Fluoxetine Collaborative Study Group. InternationalPsychogeriatrics 7, 41–53 Suppl.

Trivedi, M.H., Rush, A.J., Crismon, M.L., Kashner, T.M., Toprac, M.G., Carmody, T.J., Key, T.,Biggs, M.M., Shores-Wilson, K., Witte, B., Suppes, T., Miller, A.L., Altshuler, K.Z., Shon,S.P., 2004. Clinical results for patients with major depressive disorder in the TexasMedication Algorithm Project. Archives of General Psychiatry 61 (7), 669–680.

Trivedi, M.H., Rush, A.J., Wisniewski, S.R., Nierenberg, A.A., Warden, D., Ritz, L., Norquist,G., Howland, R.H., Lebowitz, B., McGrath, P.J., Shores-Wilson, K., Biggs, M.M.,Balasubramani, G.K., Fava, M., STAR⁎D Study Team, 2006. Evaluation of outcomeswith citalopram for depression using measurement-based care in STAR⁎D:implications for clinical practice. American Journal of Psychiatry 163 (1), 28–40.

Ulrich, G., Haug, H.J., Fahndrich, E., 1994. Acute vs. chronic EEG effects in maprotiline-and in clomipramine-treated depressive inpatients and the prediction oftherapeutic outcome. Journal of Affective Disorders 32 (3), 213–217.

Ulrich, G., Haug, H.J., Stieglitz, R.D., Fahndrich, E., 1988. Are there distinct biochemicalsubtypes of depression? EEG characteristics of clinically defined on-drugresponders and non-responders. Journal of Affective Disorders 15 (2), 181–185.

Andrew Leuchter, M.D., has provided scientific consultation or served on advisoryboards for Aspect Medical Systems, Eli Lilly and Company, Novartis Pharmaceuticals,MEDACorp, AstraZeneca, Takeda Pharmaceuticals, and Merck & Co. He has served on aspeaker's bureau for Eli Lilly and Company and Wyeth-Ayerst Pharmaceuticals. He hasreceived research/grant support from the National Institute of Mental Health, theNational Center for Complementary and Alternative Medicine, Aspect MedicalSystems, Eli Lilly and Company, Novartis Pharmaceuticals, Wyeth-Ayerst Pharmaceu-ticals, Merck & Co., Pfizer, Vivometrics, and MedAvante. He also is a minor stockholderin Aspect Medical Systems.

Ian A. Cook, M.D., has served as an advisor and consultant for Ascend Media, Bristol-Meyers Squibb, Cyberonics Inc., and Janssen. He has served on the Speaker's Bureau forBristol-Meyers Squibb, Medical Education Speakers Network, Pfizer PharmaceuticalsInc., and Wyeth Pharmaceuticals. Dr. Cook receives Research Support from AspectMedical Systems, Cyberonics Inc., Eli Lilly & Company, Novartis Pharmaceuticals, Pfizer,Inc., and Sepracor.

LaurenMarangell, M.D., currently is an employee of Eli Lilly and Company, Indianapolis,IN. Thework described in thismanuscriptwas performedwhile shewas on the faculty ofthe Baylor College of Medicine and does not necessarily reflect the views of Eli Lilly andCompany. She previously served as a consultant for, or received lecture honoraria from,Aspect Medical Systems, Cyberonics, Inc., Medtronics, GlaxoSmithKline, Pfizer, Inc.,Novartis Pharmaceuticals, and Forest Pharmaceuticals. Dr. Marangell had receivedresearch support from Bristol-Myers Squibb Company, Cyberonics, Inc., Neuronetics,National Institute of Mental Health, Stanley Foundation, NARSAD, American Foundationfor Suicide Prevention, Aspect Medical Systems, and Sanofi-Aventis.

William S. Gilmer, M.D., has served on the Speaker's Bureau for GlaxoSmithKline andPfizer. He has also received honoraria from GlaxoSmithKline and Pfizer, Inc.Additionally, Dr. Gilmer receives Research Support from Abbott, Aspect MedicalSystems, Forest Pharmaceuticals, Janssen, the National Institute of Mental Health,Neuronetics, Novartis Pharmaceuticals, and Pfizer.

131A.F. Leuchter et al. / Psychiatry Research 169 (2009) 124–131

Karl S. Burgoyne, M.D., has received Research Support from Aspect Medical Systems.

Robert H. Howland, M.D., has received Research Support from Aspect MedicalSystems, Bristol-Myers Squibb, Cederroth, Cyberonics Inc., Forest Pharmaceuticals, andNovartis Pharmaceuticals.

Madhukar H. Trivedi, M.D., has served as an advisor and consultant for AbbottLaboratories, Akzo (Organon Pharmaceuticals), Bayer, Bristol-Myers Squibb Company,Cephalon, Cyberonics, Inc., Forest Pharmaceuticals, GlaxoSmithKline, Janssen Pharma-ceutica Products, Johnson & Johnson PRD, Eli Lilly & Company, Meade Johnson, Parke-Davis Pharmaceuticals, Pfizer, Inc., Pharmacia & Upjohn, Sepracor, Solvay Pharmaceu-ticals, Inc., and Wyeth-Ayerst Laboratories. He has served on the Speaker's Bureau forAkzo (Organon Pharmaceuticals), Bristol-Myers Squibb Company, Cephalon, Inc.,Cyberonics, Inc., Forest Pharmaceuticals, Janssen Pharmaceutica Products, LP, Eli Lilly &Company, Pharmacia & Upjohn, Solvay Pharmaceuticals, Inc., and Wyeth-AyerstLaboratories. He has received Research Support from Bristol-Myers Squibb Company,Cephalon, Inc., Corcept Therapeutics, Inc., Cyberonics, Inc., Eli Lilly & Company, ForestPharmaceuticals, GlaxoSmithKline, Janssen Pharmaceutica, Merck, Novartis Pharma-ceuticals, Pfizer, Inc., Pharmacia & Upjohn, Predix Pharmaceuticals, Solvay Pharma-ceuticals, and Wyeth-Ayerst Laboratories.

Sidney Zisook, M.D., has served as an advisor and consultant for Glaxo-Smith Kline. Hehas served on the Speaker's Bureau for Glaxo-Smith Kline and Forest Laboratories.Additionally, Dr. Zisook has received Research Support from Aspect Medical Systems,PemLab, and Jed Foundation.

Rakesh Jain, M.D., M.P.H., has served as an advisor and consultant for AddrenexPharmeceuticals, Eli Lilly & Company, Forest Laboratories, Pfizer, Impax Laboratories,Shire Pharmaceuticals, and Takeda Pharmaceuticals. He has served on Speaker'sBureaus for Eli Lilly & Company, Pfizer, Inc., Shire Pharmaceuticals, and TakedaPharmaceuticals. He has received Research Support from Abbott Laboratories, AspectMedical Systems, Merck, Eli Lilly, GlaxoSmithKline, Shire Pharmaceuticals, Pfizer, andForest Pharmaceuticals.

James T. McCracken, M.D., has served as an advisor and consultant for AbbottLaboratories, Bristol-Meyers Squibb, Eli Lilly & Company, Janssen, Novartis Pharma-ceutical, and Wyeth. He has served on the Speaker's Bureau for Abbott Laboratoriesand Genentech, Inc. Dr. McCracken receives Research Support from Bristol-MeyersSquibb, Eli Lilly & Company, and Shire.

Maurizio Fava, M.D., has served as an advisor and consultant for Aspect MedicalSystems, Astra-Zeneca, Bayer AG, Biovail Pharmaceuticals, BrainCells, Inc. Bristol-Myers Squibb, Cephalon, Compellis, Cypress Pharmaceuticals, Dov Pharmaceuticals, EliLilly, EPIX Pharmaceuticals, Fabre-Kramer Pharmaceuticals, Forest Pharmaceuticals,GlaxoSmithKline, Grunenthal GmBH, Janssen Pharmaceutica, Jazz Pharmaceuticals, J &J Pharmaceuticals, Knoll Pharmaceuticals, Lundbeck, MedAvante, Neuronetics, NovartisPharmaceuticals, Nutrition 21, Organon, PamLab LLC, Pfizer, PharmaStar, Pharmavite,Roche, Sanofi/Synthelabo, Sepracor, Solvay Pharmaceuticals, Somaxon, SomersetPharmaceuticals, and Wyeth-Ayerst Laboratories. He has served on Speaker's Bureausfor Astra-Zeneca, Boehringer-Ingelheim, Bristol-Myers Squibb, Cephalon, Eli Lilly &Company, Forest Pharmaceuticals, GlaxoSmithkline, Novartis Pharmaceuticals, Orga-non, Pfizer, PharmaStar, and Wyeth-Ayerst Laboratories. He has received ResearchSupport from Abbott Laboratories, Alkermes, Aspect Medical Systems, Astra-Zeneca,Bristol-Myers Squibb, Cephalon, Eli Lilly & Company, Forest Pharmaceuticals,GlaxoSmithKline, J & J Pharmaceuticals, Lichtwer Pharma GmbH, Lorex Pharmaceu-ticals, Novartis Pharmaceuticals, Organon, PamLab LLC, Pfizer, Pharmavite, Roche,Sanofi/Synthelabo, Solvay Pharmaceuticals, and Wyeth-Ayerst Laboratories.

Dan V. Iosifescu, M.D., has provided scientific consultation to Cephalon, Inc., ForestLaboratories, Gerson Lehrman Group, and Pfizer Inc. He serves on the Speaker's Bureaufor Cephalon Inc., Eli Lilly & Company, Forest Laboratories, and Pfizer Inc. He hasreceived research/grant support from Aspect Medical Systems, Forest Laboratories, andJanssen Pharmaceutica.

Scott Greenwald, Ph.D., is an employee and a stockholder of Aspect Medical Systems, Inc.