Journal of Affective Disorders

Journal of Affective Disorders 136 (2012) 1 8 Contents lists available at SciVerse ScienceDirect Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad Research report Two-year outcomes in first-episode psychotic depression The McLean Harvard first-episode project Mauricio Tohen a,b,, Hari-Mandir K. Khalsa b, Paola Salvatore b,c, Eduard Vieta b,d, Caitlin Ravichandran e, Ross J. Baldessarini b a University of Texas Health Science Center at San Antonio, United States b Department of Psychiatry, Harvard Medical School & McLean Hospital, Belmont, MA, United States c Section of Psychiatry, Department of Neuroscience, University of Parma, Italy d Department of Psychiatry, Hospital Clinic, IDIBAPS, CIBERSAM, University of Barcelona, Spain e Laboratory for Psychiatric Biostatistics, McLean Hospital, Belmont, MA, United States article info abstract Article history: Received 17 June 2011 Received in revised form 19 August 2011 Accepted 24 August 2011 Available online 23 September 2011 Keywords: First episode Outcome Psychotic-depression Recovery Remission Relapse Objective: Early assessment can guide accurate diagnosis, prognosis, and treatment-planning for patients with major mental illnesses. Longitudinal studies in psychotic depression from onset are rare, encouraging the present study. Method: We followed 56 DSM-IV MDD patients with psychotic features prospectively and systematically to assess course and predictors of operationally-defined syndromal remission, syndromal recovery, symptomatic remission, functional recovery, and new episodes, and to evaluate diagnostic stability. Results: Among 49/56 cases followed for 2 years, 59% retained the initial diagnosis and most achieved syndromal remission (86%) and recovery (84%); 58% remitted symptomatically, and only 35% (17/49) recovered functionally. Syndromal recovery was earlier following subacute onset, lower initial depression scores, and lack of moodincongruent psychotic features. Within 2 years, 45% (22/49) experienced new episodes earlier with younger onset and higher CGI scores. DSM diagnosis changed in 41%, to bipolar (33%), or schizoaffective disorders (12%), which followed early mania-like or schizophrenia-like features, respectively. Conclusions: Within 2 years of first-hospitalizations, 41% of patients initially diagnosed with psychotic-depression met criteria for DSM-IV bipolar or schizoaffective disorders. Of the 59% retaining the initial diagnosis for 2 years, nearly half experienced new episodes, 42% remained symptomatic, and two-thirds failed to regain their own prior functional status. 2011 Elsevier B.V. All rights reserved. 1. Introduction Several studies suggest that psychotic-depression may be a distinct syndrome, but with variable uniformity in clinical features, treatment-response, longitudinal course and outcome, and of some biological measures (Schatzberg and Rothschild, 1992). Moreover, studies of the potential impact of early psychotic features on course and outcome in recurrent major Corresponding author at: University of Texas Health Science Center at San Antonio, San Antonio, TX 78229, United States. Fax: +1 210 567 7829. E-mail address: tohen@uthscsa.edu (M. Tohen). depressive disorders (MDD) are rare and inconsistent. Coryell et al. (1990) reported that MDD with versus without psychotic features had a more chronic course, more recurrences, and slower recovery from acute episodes, particularly if psychotic features were mood-incongruent. Others have found minor or insignificant differences in long-term clinical outcomes among MDD patients with and without psychotic features (Fennig et al., 1996; Jager et al., 2005; Schatzberg and Rothschild, 1992). Systematic knowledge of the course of major psychiatric disorders can contribute importantly to accurate diagnosis, prognosis, and treatment-planning (Tohen et al., 2003). Prospective studies starting near illness-onset are especially valuable, and 0165-0327/$ see front matter 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jad.2011.08.028

2 M. Tohen et al. / Journal of Affective Disorders 136 (2012) 1 8 less likely to be confounded by long-term illness, disability, or adverse effects of prolonged treatment. Since 1989, the McLean Harvard First-Episode Project has followed initially psychotic or manic patients from first-lifetime hospitalizations (Salvatore et al., 2007, 2009, 2010; Tohen et al., 1992, 2000a, 2000b, 2003). We now report on the first two years of prospective and systematic, long-term assessment of 56 firstepisode patients initially diagnosed with an episode of DSM-IV MDD with psychotic features ( psychotic-depression ). Our principal aims were to estimate rates, latencies and predictors of specific outcomes (remission, recovery, and new illness) as defined by the International Society for Bipolar Disorders (ISBD (Tohen et al., 2009)), and to determine risks of new illness-episodes and diagnostic changes. 2. Methods 2.1. Patient-subjects Research methods employed were detailed previously (Salvatore et al., 2007, 2009, 2010; Tohen et al., 1992, 2000a, 2000b, 2003). Briefly, experienced evaluators recruited and assessed subjects from inpatient units at McLean Hospital within 72 h of first-lifetime psychiatric hospitalizations in 1989 1996. Diagnoses were updated to DSM-IV criteria after 1994, and to DSM-IV-TR criteria after 2000 (American Psychiatric Association (APA)), and all subjects underwent repeated assessments between six months and several years of follow-up, including structured examinations at intake and 2 years. Stability of diagnoses over the initial two years was assessed with blinding to initial diagnoses (Leckman et al., 1982). Patients were aged 18 72 years at intake, initially met SCID-based DSM criteria (updated to DSM-IV-TR) for an acute episode of MDD with psychotic features. All subjects provided written informed consent for study procedures and for anonymous and aggregate reporting of clinical findings, following initial McLean Hospital IRB approval, with yearly reviews. Exclusion criteria were: [a] current intoxication or substance-withdrawal; [b] delirium; [c] previous psychiatric hospitalization, unless for detoxification; [d] documented IQb70, [e] ill continuously 12 months prior to intake; or [f] ever treated with an antidepressant, mood-stabilizer or antipsychotic for a total of N3 months. Inpatients were assessed approximately weekly until discharged, and again at 6 and 12 months post-discharge, and yearly thereafter. Treatment was determined clinically by clinicians not involved in the study, not influenced by study requirements, but documented during hospitalization and at follow-up assessments (Fig. 1). 2.2. Clinical assessments Baseline information collected by specifically trained research assistants included demographic and salient clinical data, assessments of psychosocial functioning in the preceding year, symptom rating-scale scores, narrative summaries of medical records, and findings of structured interviews of subjects and semi-structured interviews of patients, relatives, and treating clinicians. Structured assessments at baseline and 24 months involved Structured Clinical Interviews for DSM-Patient Version (SCID-P (Spitzer et al., 1988)) by Master's-level raters with N5 years of experience, repeatedly tested to maintain high inter-rater reliability (intra-class correlation coefficient [ICC] Syndromal remission Syndromal recovery Symptomatic remission Functional recovery 0.90). Detailed clinical narratives were prepared from all available information to support final DSM-IV diagnoses by a best-estimate procedure to achieve diagnostic consensus (Leckman et al., 1982). Symptom-ratings included: [a] the Expanded Brief Psychiatric Rating Scale (BPRS-E (Tohen et al., 2000a, 2000b; Lukoff et al., 1986)) with 36-items, including 35 individual factors for major affective and psychotic disorders and a global score, each rated for severity (absent to severe: 0 7), as well as a total summed score and sub-scale scores for psychosis, mania, and depression (Tohen et al., 1990); [b] Clinical Global Impression (CGI) scale, rated 1 7 (least to greatest (Guy, 1976)); [c] Global Assessment of Functioning (GAF) scale, scored 0 100 (worst to best (American Psychiatric Association (APA))); [d] Young Mania Rating Scale (YMRS (Young et al., 1978)); and [e] 24-item Hamilton Depression Rating Scale (HDRS (Hamilton, 1960)). Functional status was rated for occupational level with the Modified Vocational Status Index (MVSI) and residential status with a Modified Location Code Index (MLCI), scored 1 9 (both worst-to-best (Tohen et al., 1990; Dion et al., 1988)), based on information from subjects, family members and medical records; current functional status was compared to optimal status during the year prior to study-intake to assess whether individually defined residential and occupational functioning was regained (Leckman et al., 1982). 2.3. Outcome assessments and definitions 6 months 12 months 24 months 0 10 20 30 40 50 60 70 80 90 Fig. 1. Rates (%) of syndromal, symptomatic, or functional remission or recovery (as defined in the text) at 6 (gray), 12 (striped), and 24 months (black bars) among 56 first-episode patients who presented initially in major depressive episodes with psychotic features. At assessments following index hospitalizations, follow-up information (including interval and current symptomatic, occupational and residential status, and treatment) was obtained systematically by experienced raters held blind to baseline information, using semi-structured, face-to-face or telephonebased interviews (Revicki et al., 1997). Inter-rater reliability for SCID-P-based diagnoses yielded an average ICC of 0.92 for primary diagnoses, and 0.90 for secondary disorders (Woerner et al., 1988). High inter-rater reliability also was obtained for BPRS-E ratings (ICC=0.96), and between telephone and inperson outcome assessments (ICC=0.90). Syndromal remission was defined according to Frank et al. (1991) and by the International Society for Bipolar Disorders (Tohen et al., 2009), based on life-charting methods (Keller et al., 1987), as follows: no DSM-IV depression criterion scored N3, nor more than three criteria scored at 3; in addition, CGI

M. Tohen et al. / Journal of Affective Disorders 136 (2012) 1 8 3 was 3. Syndromal recovery required 8 weeks of sustained syndromal remission. Symptomatic remission was defined as scores b3 for BPRS-E-Depression subscale items (anhedonia, appetite, care, depression, dysphoria, fatigue, guilt, indecisiveness and poor concentration, motor retardation, suicidal ideation or behaviors) and for BPRS-E-Psychosis subscale items (blunted affect, thought-broadcasting, -control or -withdrawal, conceptual disorganization, disorientation, hallucinations, mannerisms or posturing, suspiciousness, unusual thought content, or emotional withdrawal (American Psychiatric Association (APA); Fekadu et al., 2009; Fennig et al., 1996)). Functional recovery required that ratings of both MVSI occupational, and MLCI residential status (Hamilton, 1960; Tohen et al., 2009) returned to or exceeded the highest individual levels within the pre-intake year, based on best-estimates from all available information (Dion et al., 1988; Leckman et al., 1982; Tohen et al., 2000a, 2000b; Woerner et al., 1988). New episodes met DSM-IV-TR criteria for a major affective, or psychotic disorder. Time-to-new-episode was estimated as days from initial syndromal recovery to the start of a new episode. Illness-onset was rated as acute ( 4weeks), subacute (5 26 weeks) or gradual (evolving over N6months).Individual BPRS-E items provided a total score and subscale scores for Depression, Mania, and Psychosis. BPRS-E total and subscale scores, CGI (6 months), Syndrome of Negative [psychotic] Symptoms (SANS (Andreasen, 1989) at baseline), Syndrome of Positive [psychotic] Symptoms (SAPS (Andreasen, 1990) at baseline), GAF (6 months), age, and days of initial hospitalization were recorded at each follow-up assessment. 2.4. Assessments of prodromal morbidity and details of index episode We also carried out comprehensive, retrospective reviews of all available documents to describe detailed clinical phenomena associated with first index episodes of psychotic-mdd, with the expert rater (PS) held blind to diagnostic outcomes. Psychopathological details of index episodes and prodromal morbidity included affective, psychotic, psychomotor, cognitive, social, and behavioral dimensions. These were supported by application of the 100-item, Manual for Assessment & Documentation of Psychopathology (AMDP-system (Ebel et al., 1989)) and the 67-item Bonn Scale for Assessment of Basic Symptoms (BABS (Guy and Ban, 1983; Klosterkötter et al., 2001)) to guide preparation of comprehensive, systematic, and detailed symptominventories. 2.5. Statistical analyses We assessed categorical variables with contingency-tables (χ 2 ), with degrees-of-freedom [df]=1, unless stated otherwise, and continuous variables with nonparametric Mann Whitney (U) rank-methods. Correlations (r s ) used the Spearman nonparametric rank method. Time (weeks) to recovery and recurrence were estimated using Kaplan Meier survival-analysis (Klein and Moeschberger, 1992). Times were censored at 24 months (or last contact) post-discharge for initial remission, recovery, and for new illness-episodes. Significance of candidate predictors of recovery, recurrence and diagnostic stability, was assessed with Mantel Cox log-rank or Fisher-exact tests (for categorical factors) or univariate Cox or logistic regression (for continuous factors). For regression models, p-values were calculated using likelihood ratio-test statistics. Goodness-of-fit was checked with Arjas plots (Klein and Moeschberger, 1992) for Cox regressions, and partial-residual plots (Hosmer and Lemeshow, 1999) for logistic regression. Logistic and linear regression, using generalized estimating equations to account for clustering within subjects, compared the probability of treatment with each drug and number of drugs used at discharge and 24-month follow-up. Statistical significance required 2-tailed pb0.05. Statistical analyses employed commercial microcomputer programs (Stata-9, Stata Corp., College Station, TX). 3. Results 3.1. Subject characteristics Subjects of investigation were recruited from an initial sample of 784 psychotic or manic patients in first-lifetime psychiatric hospitalizations (1989 1996), as detailed elsewhere (Tohen et al., 1992, 2000a, 2000b, 2003). Of the total, 64 (8.2%) had a first-episode diagnosed as MDD with psychotic features based on criteria updated to by DSM-IV-TR. Of these 64 eligible candidates, 56 (87.5%) were enrolled, including 28 men and 28 women of average age 36.3 (18 75, SD=14.7) years. Followup after initial hospital discharge averaged 3.97 (0.53 9.0, SD=2.66) years; of the 56 subjects enrolled, 49 (87.5%) were followed for 2 years; 6 were lost at 12 months. There were no clear differences in sex, age, or initial illness-severity ratings between those recruited or not, followed or lost (not shown), although the numbers involved are small. Most patients were Caucasian, completed high school, were unmarried, and psychiatric, neurological, and general medical co-morbidities were prevalent (Table 1). Co-morbid conditions ranked: alcohol or drug abuse or dependence (89.3%)Ngeneral medical disorders (66.1%)Nneuromedical conditions (50.1%)= an anxiety or post-traumatic stress disorder(50.0%;table 1). Most patients (76.8%) showed a sub-acute or gradual onset. Average initial ratings on standardized scales indicated moderate or marked morbidity, except for expected low YMRS mania ratings (Table 1). Initial hospitalizations averaged 25.3 (SD=22.1)days (Table 1), but this measure declined markedly across the years of subject-recruitment, from 43.7 (SD=16.0)days in 1989 to only 12.7 (SD=4.7)days in 1996 (r s = 0.566, p=0.001). The decrease in hospital stay evidently represented the impact of administrative health policy changes at the study site since major measures of initial morbidity were unrelated to year at intake (e.g., for total BPRS-E, r s =0.11, p=0.42). 3.2. Diagnostic stability Among the 49 subjects followed at least 2 years, 20 (40.8%) changed diagnosis, and the diagnosis of MDD with psychotic features was retained in 29 (59.2%). Most of the 20 changed diagnoses were to a DSM-IV bipolar disorder (type I or NOS, 70.0% [14/20]) or a schizoaffective disorder (30.0% [6/20]). Subjects with changed diagnoses differed little in most measures from those who continued to be diagnosed with psychotic-mdd, except (in univariate analysis), for having: [a] 1.23 2.43-times higher initial BPRS-E-mania or YMRS scores, [b] 2.16-times less medical co-morbidity, and [c] 1.72-times more initial

4 M. Tohen et al. / Journal of Affective Disorders 136 (2012) 1 8 Table 1 Characteristics of first-episode patients initially diagnosed with DSM-IV major depressive disorder with psychotic features. Characteristics At intake (N=56) Two-year outcome χ 2 p-value Dx stable (n=29) Dx changed (n=20) Men (%) 50.0 62.1 40.0 2.33 0.13 Age at admission 36.3±14.7 38.7±16.6 33.0±12.8 1.77 0.18 Non-Caucasian (%) 14.1 6.90 20.0 1.86 0.17 Married (%) 23.2 24.1 10 1.68 0.20 Educated 12 years (%) 89.3 82.8 95 1.83 0.18 Syndrome-onset age 35.9±14.7 38.3±16.7 32.5±12.7 1.78 0.19 Onset type (%) 1.42 0.23 Acute (b1 month) 23.2 13.8 25.0 Subacute (1 6 months) 46.4 55.2 40.0 Gradual (N6 months) 30.4 31.0 35.0 Psychotic features at intake (%) 4.30 0.04 Mood-congruent 51.8 55.2 45.0 Mood-incongruent 48.2 37.9 65.0 Index days in hospital 25.3±22.1 28.9±21.7 23.0±24.4 0.86 0.35 Suicide attempt 3 months (%) 32.1 34.5 20 1.25 0.26 Psychiatric co-morbidity (%) Any Axis I (not SUD) 37.5 31.0 50.0 1.79 0.18 Anxiety disorder 35.7 31.0 40.0 0.42 0.52 PTSD 14.3 6.90 25.0 3.14 0.08 Axis II 28.6 34.5 20.0 1.25 0.26 Alcohol abuse/dependence 62.5 65.5 65.0 0.01 0.97 Drug abuse/dependence 26.8 24.1 40.0 1.40 0.24 Eating disorder 5.40 3.45 10.0 0.87 0.35 Neuromedical co-mobidity (%) Learning disability 17.9 20.7 20.0 0.01 0.95 Head trauma 17.9 13.8 20.0 0.33 0.57 Migraine 7.14 13.8 0.00 0.00 1.00 Epilepsy 7.14 6.90 10.0 0.15 0.70 General medical co-morbidity (%) Any disorder 66.1 86.2 40.0 9.68 0.002 Allergies 10.7 6.90 5.00 0.08 0.78 Initial rating-scale scores BPRS-E, Total 85.4 ±23.0 82.1 ±20.7 91.2 ±27.6 1.70 0.19 BPRS-E Depression 32.7±12.2 32.0±10.3 33.3±15.8 0.12 0.73 BPRS-E Psychosis 19.6±7.2 19.0±6.70 22.2±7.43 2.44 0.12 BPRS-E Mania 17.3±5.90 15.4±5.02 19.0±5.99 4.87 0.03 HDRS-17 23.6 ±8.20 22.5 ±7.10 24.1 ±9.90 0.28 0.60 YMRS 4.80 ±4.80 2.10 ±3.40 5.10 ±5.10 5.25 0.02 SANS 33.3 ±24.4 30.4 ±25.8 40.2 ±23.4 1.64 0.20 SAPS 23.3 ±22.4 20.5 ±15.7 29.7 ±30.0 1.59 0.21 GAF (6 months) 62.0 ±17.3 63.3 ±17.7 61.9 ±16.5 0.07 0.78 CGI (6 months) 3.24 ±1.14 3.20 ±1.12 3.20 ±1.15 0.00 1.00 Subjects with stable versus unstable diagnoses of MDD at two years were compared with multivariate logistic regression to compute likelihood ratio (χ 2 ). Rating scales are defined in Methods. Both mood-congruent and incongruent psychotic features were found in 5/56 subjects (8.9%), and mood-opposite psychotic features in 4/56 (7.14%) of subjects of which 3/4 (75%) changed diagnosis at two years. mood-incongruent psychotic features (Table 1). By multivariate logistic regression modeling of these and other selected factors of interest (not shown), those remaining independently and significantly associated with diagnostic instability by 2 years, were: [a] lower prevalence of initial medical co-morbidity (OR=11.2 [CI=2.23 56.0], p=0.003); [b] higher intake BPRS-mania score (OR=0.83 [CI=0.73 0.94], p=0.004); and [c] male sex (OR=5.43 [CI=1.24 23.7], p=0.024). 3.3. Features predicting diagnostic change Based on the analyses of psychopathological features in firstepisodes of psychotic-mdd, four factors were significantly associated with future diagnostic changes. Changes to BPD were associated with: [a] mood-lability (later BPD: 87.5% vs. not: 56.1%; χ 2 =4.85,p=0.03),and[b]subsyndromalhypomanic symptoms (62.5% vs. 30.8%; χ 2 =4.76,p=0.03).Associated withschizoaffective disorders were: [a] presence of any Schneiderian firstrank psychotic symptoms (later schizoaffective: 88.9% vs. not: 43.5%; χ 2 =6.21, p=0.01), and [b] presence of first-episode formal thought disorder (77.8% vs. 41.3%; χ 2 =4.01, p=0.04). 3.4. Clinical outcomes by two years Most subjects (85.7% [48/56]) initially diagnosed with DSM- IV psychotic depression achieved syndromal recovery within 2 years of follow-up. Median latency to syndromal recovery averaged 10.9 [95% CI: 7.10 15.4]weeks (Table 2). Those who retained the initial diagnosis of psychotic depression or changed within 2 years differed little in rates of syndromal recovery, or in median latency to it (Table 2). However, Cox multivariate regression modeling found slower syndromal recovery to be associated

M. Tohen et al. / Journal of Affective Disorders 136 (2012) 1 8 5 Table 2 Recovery, new episodes, and diagnostic changes within two years of hospitalization for first-episodes of DSM-IV major depression with psychotic features. Outcome From intake (N=56) Dx stable (n=29) Dx changed (n=20) Syndromal remission 6 months (%) 73.2 75.9 80.0 0.12 0.90 12 months (%) 83.9 96.6 80.0 3.54 0.14 24 months (%) 85.7 100.0 95.0 1.48 0.41 Median latency (weeks) 10.9 [7.1 15.4] 8.70 [5.60 17.3] 13.0 [3.00 16.4] 1.55 0.22 Syndromal recovery 6 months (%) 71.4 75.9 84.2 0.49 0.72 12 months (%) 82.1 96.6 84.2 2.29 0.29 24 months (%) 83.9 100.0 84.2 4.88 0.056 Median Latency (weeks) 10.9 [7.10 13.4] 8.70 [5.60 17.3] 13.0 [3.00 16.7] 2.11 0.15 Symptomatic remission (%) 6 months 45.5 24.1 57.1 0.33 0.57 12 months 56.7 31.0 61.5 0.20 0.65 24 months 58.3 31.0 50.0 0.01 0.91 Functional recovery (%) 6 months (%) 22.0 20.7 15.0 0.39 0.53 12 months (%) 31.9 41.4 15.0 4.86 0.028 24 months (%) 34.9 34.5 a 26.3 1.12 0.29 New episodes Depression relapse (%) 1.96 3.45 0.00 0.70 0.25 Depression recurrence (%) 33.3 37.9 25.0 0.90 0.38 All new episodes (%) 45.1 41.4 50.0 0.90 0.38 Median latency (weeks) 57.3 [39.3 75.4] N96 N96 0.36 0.57 Follow-up was for 2 years from intake at index episode; new episodes follow syndromal recovery. Dx = diagnosis. Median latency (weeks to 50% of subjects attaining recovery or experiencing a new illness-episode, with 95% CI) is based on survival analysis. Statistics are likelihood-ratio χ 2 values based on logistic regression to compare subjects with stable or changed diagnoses of MDD at 2 years. a Within 2 years: 9/29 (31.0%) of subjects gained and lost functional recovery. Statistic p-value with: [a] higher initial BPRS-E scores (total, depressive, and psychotic), [b] higher 6-month CGI score, [c] lower 6-month GAF score, and [d] history of head trauma (Table 3). However, timeto-syndromal-recovery was not associated with type of illnessonset, initial mood-incongruent psychotic features, or treatments used during initial hospitalization. Sustained syndromal recovery (remission sustained for 8 weeks) was attained by 83.9% of the 49 subjects followed to 2 years, with a median latency of 10.9 (CI: 7.10 13.4)weeks (Table 2), and with a slightly and nonsignificantly longer delay among subjects with changed diagnoses (Table 2). As expected, symptomatic remission assessed cross-sectionally at intervals up to 24 months, was achieved less often (58.3%) than syndromal remission (85.7%) or recovery (83.9%) within two years, and showed little difference between subjects with stable versus changed diagnoses (Table 2). Furthermore, functional recovery was least likely to be achieved within 2 years Table 3 Factors associated with longer time to syndromal recovery. Factors Hazard ratio [95% CI] p-value Higher 6-months CGI score 1.69 [1.27 2.27] b0.0001 Higher initial BPRS-E total score 1.03 [1.01 1.04] b0.0001 Higher initial BPRS-E-mania score 1.08 [1.03 1.14] 0.002 Lower 6-month GAF score 1.03 [1.01 1.06] 0.002 Higher initial BPRS-E-depression score 1.04 [1.01 1.06] 0.004 Higher initial BPRS-E-psychosis score 2.17 [1.03 4.55] 0.04 History of head trauma 2.17 [1.03 4.56] 0.04 Other factors found not significantly related to time to syndromal recovery included type of initial illness-onset, mood-congruence/incongruence of initial psychotic features, and type of treatment during initial hospitalization. (34.9%); functional recovery was similarly unlikely with stable or changed diagnoses (Table 2) and was not significantly associated with any factor listed in Table 1 (not shown). New episodes of depressive illness (relapse or recurrence following initial recovery) occurred among 45.1% of 49 subjects within two years, with little difference among subjects with changed or stable diagnoses (Table 2). The median latency to a new episode of illness following initial recovery was 57.3 (CI: 39.3 75.4)weeks (Table 2). Cox multivariate regression analyses of this latency indicated that only a higher 6-month CGI score (OR=1.86 [CI: 1.10 3.14]) and younger onset (OR=1.03 [CI: 1.00 1.08]) differentiated subjects with a new illness-episode from those without. 3.5. Treatment Treatments given at initial hospital-discharge and in followup were decided clinically by independent primary clinicians, not guided by the study protocol, and varied widely and changed over time, as expected (Table 4). Most subjects (93.0%) received at least one psychotropic agent at discharge, but at 2-years, 51.0% (25/49) received none. Non-treatment at 2-years was not associated with having attained syndromal recovery or symptomatic remission (both χ 2 0.411, both pn0.31). At hospital discharge, the most commonly prescribed drugs (alone or in various combinations) were antidepressants (43/56 cases: 76.8%), as expected, with less prevalent treatments ranking: sedative-anxiolytics (35.7%)Nmood-stabilizers (16.1%)Nantipsychotics (10.7%). At 2- year follow-up, 49.0% (24/49) of patients were receiving at least one psychotropic drug (Table 4). At 2 years, antidepressants were being used by 38.8% (19/49) of subjects followed that long and they remained the most frequently prescribed

6 M. Tohen et al. / Journal of Affective Disorders 136 (2012) 1 8 Table 4 Treatments given at hospital discharge and 24-month follow-up. Treatments (%) Discharge Two-years Antidepressants 76.8 38.8 Antipsychotics 10.7 22.4 Mood-stabilizers Lithium 10.71 8.16 Divalproex 5.36 12.2 Carbamazepine 0.00 2.04 Sedative-anxiolytics 35.7 18.4 Anticholinergics 32.1 2.04 Psychotropics/patient a 2.45 ±1.22 1.35 ±1.70 None 7.14 51.0 1 12.5 12.2 2 32.1 12.2 3 48.2 24.5 Monotherapies at discharge and follow-up involved an antidepressant twice as often as a mood-stabilizer. Proportions do not all add to 100% owing to use of more than one drug at a time in some patients. a The number of psychotropic drugs/patient at 2-year follow-up (excluding anticholinergics given with antipsychotics) was 45% lower than at discharge (χ 2 =16.1, p=0.0001, based on linear regression for drugcount, adjusted for clustering within subjects using generalized estimating equations), including those no longer treated. At discharge n =56, at follow-up n =49. Note that polytherapy ( 2 psychotropics/patient) involved 80.5% of patients at discharge and 36.7% at two-year follow-up. psychotropics, followed by antipsychotics (22.4% [11/49]) and mood-stabilizers (22.4% [11/49]) more than sedative-anxiolytics (18.4% [9/49]). 4. Discussion This study within the McLean Harvard First-Episode Project involved a single-site cohort of 56 hospitalized patients with an initial DSM-IV diagnosis of major depression with psychotic features (psychotic-depression) who were followed prospectively for nearly four years, with only 12.5% drop-out by two years. A particularly striking finding was that 40.7% of patients changed to other DSM-IV diagnoses by two years of follow-up, shifting to either bipolar (28.5%) or schizoaffective (12.2%) disorders. Factors associated with change of diagnosis included absence of medical co-morbidity, female sex, and early psychopathological features (bipolarlike or schizophrenialike) suggestive the final diagnoses. Since the most common new diagnosis was bipolar disorder, it may not be surprising that initial manic symptoms were found, though of relatively low intensity, insufficient to meet DSM-IV criteria for a mixed episode. Patients with an unstable diagnosis had a 2.4-fold higher baseline YMRS ratings of manic-symptoms (2.12 [SD 3.40] versus 5.05 [SD 5.09]; Odds Ratio [OR with 95% CI]: 1.18 [1.01 1.38], p=0.036). DelBello et al. (2003) previously reported a 13% switch to mania in a 3-site psychotic depression cohort that included some of the patients in the current report. These observations suggest that the proposed change in DSM 5 (American Psychiatric Association) of mixed features specifier may be useful, as our findings suggest that the presence of mixed features had implications for diagnostic outcomes. Also, and again not surprisingly, the presence of Schneiderian firstrank (mood incongruent) symptoms and prodromal thought disorder predicted diagnostic change to schizoaffective disorder (Table 1). Moreover, mood-incongruent psychotic features have been reported to predict poor outcomes in major depression (Coryell et al., 1990) as well as in bipolar-i disorder(tohen et al., 1992). The observed association of diagnostic stability with medical co-morbidity is not readily explained (Sachdeva, 2009). We recently reported on the two-year diagnostic stability of various psychotic and major affective disorders among more than 500 first-episode patients diagnosed by either DSM-IV (Salvatore et al., 2009) or ICD-10 criteria (Salvatore et al., 2010). In those studies, psychotic depression was found to be a far less stable diagnosis than bipolar-i disorder or schizophrenia, though not as unstable as acute, nonaffective psychotic disorders. It is striking that none of the present patients with final BPD diagnoses met DSM-IV criteria for BP-II disorder (all were BP-I or NOS). Lack of BP-II outcomes may reflect the samplesize, but also raises the possibility that psychotic features in a depressive episode may limit the likelihood of switching to BP-II diagnoses, even though psychotic features have been reported in a minority of bipolar-ii disorder patients (Mazzarini et al., 2010). In general, the present findings suggest that a first episode of DSM-IV psychotic-mdd might best be considered a provisional diagnosis, in need of ongoing reconsideration and refinement. The results also support the utility of the concept of mixed features specifier, even of moderate intensity or duration. Such a broadened concept of mixed manic-depressive states (Salvatore et al., 2002), in contrast to the DSM-IV requirement for meeting full criteria for mania and major depression is more consistent with ICD-10 definitions of mixed-states, and may predict later evolution of BPDs among patients initially diagnosed as psychotic-mdd by DSM-IV criteria. Strikingly, nearly all of the 29 patients with a stable two-year diagnosis of psychotic-mdd suffered from at least one co-morbid condition, including high rates of general medical, neuromedical, psychiatric, and substance-use disorders (Table 1). Such risks may be similar to, or even greater than, among bipolar-i disorder patients who have high rates of co-morbid disorders (Bowden, 2010; Gentil, 2009; Parker, 2010). Notably, we reported (Tohen et al., 2003) that, among first-episode mania patients enrolled at the same study-site, the prevalence of substance-use disorders was 18.7%, or nearly five-times less than the 89.3% observed in the present study of psychotic-mdd patients. Among psychotic-mdd patients with a stable diagnosis, general medical conditions were more than twice as prevalent as in those with changed 2-year diagnoses (66.1% versus 31.3%; Table 1). These observations again suggest that psychotic-mdd differs substantially from BP-I patients as well as from nonpsychotic MDD patients, and that the disorder requires further study. Of the present subjects with psychotic-depression, 84% achieved syndromal remission, whereas 42% remained at least moderately symptomatic, and only 35% returned to their own baseline psychosocial functioning. These findings accord with other studies of psychotic-mdd patients, who showed more severe and functional impairment over several years of follow-up than non-psychotic MDD patients, as well as particularly severe cognitive impairments (Bora et al., 2010; Coryell et al., 1990; Reichenberg et al., 2009; Schatzberg and Rothschild, 1992). New episodes (relapses or recurrences) among the present patients occurred earlier following earlier onset-age and higher CGI scores. Younger illness-onset has been identified as a predictor of poor outcome (Post et al., 2010). Our finding of relapse/recurrence in 45% of psychotic-mdd patients within 2 years is similar to the rate of 43% in 2 years reported

M. Tohen et al. / Journal of Affective Disorders 136 (2012) 1 8 7 by Naz et al. (2007). Since 91% of patients with new illness-episodes had psychotic features during follow-up, it may well be that psychotic-depression should be considered a distinct syndrome to be distinguished from nonpsychotic, and generally less severe, forms of MDD (Rothschild et al., 2008; Schatzberg and Rothschild, 1992). Treatment was not controlled or guided by the present study protocol, but reflected community clinical standards and was determined independently by each patient's prescribing clinician. Since almost all of the present patients received psychotropic medicines (usually antidepressants, and less often mood-stabilizers or antipsychotics, in various combinations) at initial hospital-discharge, it was particularly striking that more than half of the patients were no longer receiving a psychotropic drug at two years of follow-up. This loss of sustained treatment was similar to our previous findings among first-episode BP-I patients (Tohen et al., 2003), and BPD patients in community samples (Baldessarini et al., 2007, 2008). This pattern may reflect reluctance of mood-disorder patients to continue maintenance treatments when they are relatively well, or due to other factors, including adverse effects and costs of long-term treatment. However, we found that earlier syndromal recovery or symptomtic remission was not associated with treatment status at two years. 5. Limitations Limitations of the present study include moderate statistical power owing to relatively small numbers of subjects, and possible sampling bias related to requiring initial hospitalization. Also, it was not feasible to test for treatment-associated effects owing to the highly heterogeneous, changing, and uncontrolled clinical treatments involved. In addition the follow up of 2 years may not be long enough to allow a full assessment of long term diagnostic stability. It is possible that a longer follow may provide an even larger diagnostic stability. Furthermore some of the diagnostic entities such as schizoaffective disorder under the current definitions may be unreliable. Finally, the intermittent follow-up assessments may have missed some important features of intervening morbidity. In conclusion, the present findings indicate that clinical outcome in first-episode, DSM-IV major depression patients with initial psychotic features was highly variable and often followed by changes of diagnosis to bipolar or schizoaffective disorders within two years. In addition, morbid risks were high: only 58% of patients achieved symptomatic remission, and only 35% attained functional recovery to their own premorbid employment and residential status. From a clinical perspective our findings suggest that when clinicians manage the treatment of patients with a first episode of psychotic depression, they need to reassess the diagnosis within the first two years in order to determine the optimal treatment. Further assessment of similar patients should test the hypothesis that psychotic-depression may be a discrete syndrome with different prognostic and therapeutic implications compared to nonpsychotic-mdd patients. Our findings also support the proposed DSM 5 (American Psychiatric Association) concept of mixed features specifier, as their presence predicted diagnostic switch to BPD. Identification of mixed features in patients with psychotic depression may have important implications in terms of outcome and treatment management. Further research might well consider early indicators of diagnostic instability, and early preventative pharmacological strategies that can improve functional outcome. Role of funding source Supported in part by research grants from NIMH grants MH-04844, MH-10948 and the Atlas Foundation; (to MT), NARSAD (to PS), the Kenneth Rossano Research Fellowship (to H-MKK), the Spanish Ministry of Health, Instituto de Salud Carlos III, CIBERSAM, and Spanish Ministry of Education & Science (to EV), by NIH grants MH-47370 and MH-073049 and grants from the Bruce J. Anderson Foundation and the McLean Private Donors Research Fund (to RJB). Conflict of interest Dr. Tohen was formerly employed by Eli Lilly Corporation (to 2008) and has received honoraria or consulted for AstraZeneca, BristolMyersSquibb, Glaxo-SmithKline, Forest, Eli Lilly, Johnson & Johnson, Merck, Otsuka, Sepracor, Sunovion, Lundbeck and Wyeth Corporations; his spouse is a current employee and minor stockholder at Eli Lilly. Dr. Vieta is a consultant or research collaborator with Almirall, Astra-Zeneca, BristolMyersSquibb, Forest, Glaxo-SmithKline, Janssen, Lilly, Lundbeck, Merck, Novartis, Organon, Otsuka, Pfizer, Sanofi- Aventis, Servier, Takeda, and United BioSource Corporations. Drs. Salvatore, Ravichandran and Baldessarini and Ms. Khalsa and members of their immediate families have no current industrial relationships that might present potential conflicts of interest in this work. References American Psychiatric Association, n. DSM-5 Development: http://www. dsm5.org/proposedrevisions/pages/mooddisorders.aspx. American Psychiatric Association (APA). Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV) and text-revision (DSM-IV- TR). Washington, DC: APA Press, 1994 and 2000. Andreasen, N.C., 1989. Scale for Assessment of Negative Symptoms (SANS): conceptual and theoretical foundations. Br. J. Psychiatry 154 (7 Suppl.), 49 58. Andreasen, N.C., 1990. Methods for assessing positive and negative symptoms. Mod. Probl. Pharmacopsychiatry 24, 73 88. Baldessarini, R.J., Leahy, L.F., Arcona, S., Gause, D., Zhang, W., Hennen, J., 2007. Prescribing patterns of psychotropic medicines in the United States for patients diagnosed with bipolar disorders. Psychiatr. Serv. 58, 85 91. Baldessarini, R.J., Henk, H.J., Sklar, A.R., Chang, J., Leahy, L.F., 2008. Use of psychotropic medicines in bipolar disorder patients in the United States. Psychiatr. Serv. 59, 1175 1183. Bora, E., Yücel, M., Pantelis, C., 2010. Cognitive impairment in schizophrenia and affective psychoses: implications for DSM-V criteria and beyond. Schizophr. Bull. 36, 36 42. Bowden, C.L., 2010. Comorbidities with bipolar disorders: significant, recognition, and management. CNS Spectr. 15 (Suppl. 2), 8 9. Coryell, W., Keller, M., Lavori, P., Endicott, J., 1990. Affective syndromes, psychotic features, and prognosis. Arch. Gen. Psychiatry 47, 658 662. DelBello, M.P., Carlson, G.A., Tohen, M., Bromet, E.J., Schwiers, M., Strakowski, S.M., 2003. Rates and predictors of developing a manic or hypomanic episode 1 to 2 years following a first hospitalization for major depression with psychotic features. J. Child Adolesc. Psychopharmacol. 13 (2), 173 185 Summer. Dion, G.L., Tohen, M., Anthony, W.A., Waternaux, C.S., 1988. Symptoms and functioning of patients with bipolar disorder six months after hospitalization. Hosp Community Psychiatry 39, 652 657. Ebel, H., Gross, G., Klosterkötter, J., Huber, G., 1989. Basic symptoms in schizophrenic and affective psychoses. Psychopathology 22, 224 232. Fekadu, A., Wooderson, S.C., Markopoulo, K., Donaldson, C., Papadopoulos, A., Cleare, A.J., 2009. What happens to patients with treatment-resistant depression? A systematic review of medium to long term outcome studies. J. Affect. Disord. 116, 4 11. Fennig, S., Bromet, E.J., Karant, M.T., Ram, R., Jandorf, L., 1996. Mood-congruent vs. mood-incongruent psychotic symptoms in first-admission patients with affective disorder. J. Affect. Disord. 37, 23 29. Frank,E.,Prien,R.F.,Jarrett,J.B.,Keller,M.B.,Kupfer,D.J.,Lavori,P.W.,Rush,A.J., Weissman, M.M., 1991. Conceptualization and rationale for consensus definitions of terms in major depressive disorder: response, remission, recovery, relapse, and recurrence. Arch. Gen. Psychiatry 48, 851 855. Gentil, V., 2009. Bipolar disorder and concurrent psychiatric and medical disorders. J. Clin. Psychiatry 70, e29.

8 M. Tohen et al. / Journal of Affective Disorders 136 (2012) 1 8 ECDEU Assessment Manual for Psychopharmacology: Publication ABM 76-338. In: Guy, W. (Ed.), Education and Welfare. US Department of Health, Washington DC, pp. 218 222. Guy, W., Ban, T.A., 1983. The AMDP and NCDEU/BLIPS systems: similarities and differences. Mod. Probl. Pharmacopsychiatry 20, 185 192. Hamilton, M.A., 1960. Rating scale for depression. J. Neurol. Neurosurg. Psychiatry 23, 56 62. Hosmer, D.W., Lemeshow, S., 1999. Applied Logistic Regression. Wiley, New York, pp. 49 52. Jager, M., Bottlender, R., Strauss, A., Moller, H.J., 2005. Fifteen-year follow-up of DSM-IV depressive disorders: prognostic significance of psychotic features. Compr. Psychiatry 46, 322 327. Keller, M.B., Lavori, P.W., Friedman, B., Nielsen, E., Endicott, J., McDonald- Scott, P., Andreasen, N.C., 1987. The longitudinal interval follow-up evaluation: comprehensive method for assessing outcome in prospective longitudinal studies. Arch. Gen. Psychiatry 44, 540 548. Klein, J.P., Moeschberger, M.L., 1992. Survival Analysis: Techniques for Censored and Truncated Data. Springer, New York. Klosterkötter, J., Hellmich, M., Steinmeyer, E.M., Schultze-Lutter, F., 2001. Diagnosing schizophrenia in the initial prodromal phase. Arch. Gen. Psychiatry 58, 158 164. Leckman, J.F., Sholomskas, D., Thompson, W.D., Belanger, A., Weissman, M.M., 1982. Best-estimate of lifetime psychiatric diagnosis: methodological study. Arch. Gen. Psychiatry 39, 879 883. Lukoff, D., Neuchterlein, K.H., Ventura, J., 1986. Manual for the expanded brief psychiatric rating scale. Schizophrenia Bull 112, 594 602. Mazzarini, L., Colom, F., Pacchiarotti, I., Nivoli, A.M., Murru, A., Bonnin, C.M., Cruz, N., Sanchez-Moreno, J., Kotzalidis, G.D., Girardi, P., Tatarelli, R., Vieta, E., 2010. Psychotic versus non-psychotic bipolar II disorder. J. Affect. Disord. 126, 55 60. Naz, B., Craig, T.J., Bromet, E.J., Finch, S.J., Fochtmann, L.J., Carlson, G.A., 2007. Remission and relapse after the first hospital admission in psychotic depression: four-year naturalistic follow-up. Psychol. Med. 37, 1173 1181. Parker, G.B., 2010. Comorbidities in bipolar disorder: models and management. Med. J. Aust. 193 (Suppl. 4), S18 S20. Post, R.M., Leverich, G.S., Kupka, R.W., Keck Jr., P.E., McElroy, S.L., Altshuler, L.L., Frye, M.A., Luckenbaugh, D.A., Rowe, M., Grunze, H., Suppes, T., Nolen, W.A., 2010. Early-onset bipolar disorder and treatment delay are risk factors for poor outcome in adulthood. J. Clin. Psychiatry 71, 864 872. Reichenberg, A., Harvey, P.D., Bowie, C.R., Mojtabai, R., Rabinowitz, J., Heaton, R.J., Bromet, E., 2009. Neuropsychological function and dysfunction in schizophrenia and psychotic affective disorders. Schizophrenia Bull 35, 1022 1029. Revicki, D.A., Tohen, M., Gyulai, L., Thompson, C., Pike, S., Davis-Vogel, A., Zarate, C., 1997. Telephone vs. in-person clinical and health status assessment interviews in patients with bipolar disorder. Harv. Rev. Psychiatry 5, 75 81. Rothschild, A.J., Winer, J., Flint, A.J., Mulsant, B.H., Whyte, E.M., Heo, M., Fratoni, S., Gabriele, M., Kasapinovic, S., Meyers, B.S., 2008. Study of Pharmacotherapy of Psychotic Depression (STOP-PD) Collaborative Study Group. Missed diagnosis of psychotic depression at 4 academic medical centers. J. Clin. Psychiatry 69, 1293 1296. Sachdeva, S., 2009. Depression and chronic medical illness: why psychiatrists should care. Psychopharmacology Educational Update. Neuroscience Education Institute http:/www.neiglobal.com/allarticles.aspx?categoryid= 3&contented. [accessed 10 June 2009]. Salvatore, P., Baldessarini, R.J., Centorrino, F., Egli, S., Albert, M., Gerhard, A., Maggini, C., 2002. Weygandt's, the manic-depressive mixed states: a translation and commentary on its significance in the evolution of the concept of bipolar manic-depressive disorder. Harv. Rev. Psychiatry 10, 255 275. Salvatore, P., Tohen, M., Khalsa, H.M., Baethge, C., Tondo, L., Baldessarini, R.J., 2007. Longitudinal research on bipolar disorders. Epidemiol. Psichiatr. Soc. 16, 109 117. Salvatore, P., Baldessarini, R.J., Tohen, M., Khalsa, H.M., Sanchez-Toledo, J.P., Zarate Jr., C.A., Vieta, E., Maggini, C., 2009. McLean Harvard International First-Episode Project: two-year stability of DSM-IV diagnoses in 500 first-episode psychotic disorder patients. J. Clin. Psychiatry 70, 458 466. Salvatore, P., Khalsa, H.M.K., Perez Sanchez-Toledo, J., Tohen, M., Maggini, C., Baldessarini, R.J., 2011. Stability of initial ICD-10 diagnoses in 443 firstepisode psychotic disorder subjects. J. Clin. Psychiatry 72 (2), 183 193. Schatzberg, A.F., Rothschild, A.J., 1992. Psychotic (delusional) major depression: should it be included as a distinct syndrome in DSM-IV? Am. J. Psychiatry 149, 733 745. Spitzer, R.L., Williams, J.B.W., Gibbon, M., First, M.B., 1988. Structured Clinical Interview for DSM-III-R Patient Version (SCID-P). New York State Psychiatric Institute, Biometrics Research Department, New York. Tohen, M., Waternaux, C.M., Tsuang, M.T., 1990. Outcome in mania: fouryear prospective follow-up of 75 patients utilizing survival analysis. Arch. Gen. Psychiatry 47, 1106 1111. Tohen,M.,Stoll,A.L.,Strakowski,S.M.,Faedda,G.L.,Mayer,P.V.,Goodwin,D.C., Kolbrener, M.L., Madigan, A.M., 1992. McLean first-episode psychosis project: six month recovery and recurrence outcome. Schizophrenia Bull 18, 172 185. Tohen, M., Hennen, J., Zarate Jr., C.A., Baldessarini, R.J., Strakowski, S.M., Stoll, A.L., Faedda, G.L., Suppes, T., Gebre-Medhin, P., Cohen, B.M., 2000a. Twoyear syndromal and functional recovery in 219 cases of major affective disorders with psychotic features. Am. J. Psychiatry 157, 220 228. Tohen,M.,Strakowski,S.M.,Hennen,J.,ZarateJr.,C.A.,Stoll,A.L.,Suppes,T.,Faedda, G.L., Cohen, B.M., Gebre-Medhin, P., Baldessarini, R.J., 2000b. McLean Harvard first-episode project: six-month symptomatic and functional outcome in affective and nonaffective psychoses. Biol. Psychiatry 48, 467 476. Tohen, M., Zarate Jr., C.A., Hennen, J., Khalsa, H.M., Strakowski, S.M., Gebre- Medhin, P., Salvatore, P., Baldessarini, R.J., 2003. The McLean Harvard first-episode mania study: prediction of recovery and first recurrence. Am. J. Psychiatry 160, 2099 2107. Tohen, M., Frank, E., Bowden, C.L., Colom, F., Ghaemi, N.S., Yatham, L.N., Malhi, G.S., Calabrese, J.R., Nolen, W.A., Vieta, E., Kapczinski, F., Goodwin, G.M., Suppes, T., Sachs, G.S., Chengappa, K.N.R., Grunze, H., Mitchell, P.B., Kanba, S., Berk, M., 2009. The International Society for Bipolar Disorders (ISBD) Task Force report on the nomenclature of course and outcome in bipolar disorders. Bipolar Disord. 11, 453 473. Woerner, M.G., Manuzza, S., Kane, J.M., 1988. Anchoring the BPRS: an aid to improved reliability. Psychopharmacol. Bull. 24, 112 117. Young, R.C., Biggs, J.T., Ziegler, V.E., Meyer, D.A., 1978. A rating scale for mania: reliability, validity and sensitivity. Br. J. Psychiatry 133, 429 435.