|Home | About | Journals | Submit | Contact Us | Français|
The aim of this study was to examine predictors of diagnostic and symptomatic outcome in adolescents with either psychotic disorder not otherwise specified (PsyNOS) or brief psychotic disorder (BrPsy) followed in a schizophrenia prodromal program.
As part of a naturalistic study of adolescents considered at clinical high risk for schizophrenia, 26 youths (mean age, 15.9±2.6 years, 65.4% male) with psychosis not fulfilling criteria for schizophrenia/schizoaffective disorder and diagnosed with PsyNOS or BrPsy were evaluated for predictors of diagnostic and symptomatic outcome after at least 6 (mean, 22.8±19.4) months follow up.
Progression to schizophrenia, schizoaffective disorder, or psychotic bipolar disorder (n=10, 38.5%) was predicted by fulfilling criteria for schizotypal personality disorder at baseline (p =0.046). Development of schizophrenia/schizoaffective disorder (n=7, 27.0%) was associated with worse executive functioning (p=0.029) and absence of anxiety disorders (p=0.027). Conversely, progression to bipolar disorder (n=4, 15.4%), with (n=3, 11.5%) or without (n=1, 3.8%) psychosis, was associated with the presence of anxiety disorders (p=0.014). Remission of all psychotic as well as attenuated positive or negative symptoms (n=5, 19.4%) was predicted by Hispanic ethnicity (p=0.0047), an initial diagnosis of BrPsy (p=0.014), longer duration of antidepressant treatment (p=0.035), and better attention at baseline (p=0.042).
Results from this preliminary study suggest that patients with PsyNOS, BrPsy, or schizotypal personality disorder features in adolescence should be followed as separate risk groups in prodromal studies of schizophrenia and bipolar disorder. Executive function deficits and absence of anxiety disorders may be risk markers for schizophrenia, while presence of anxiety disorders may be linked to bipolar disorder risk. After achieving full remission, patients with sudden onset of psychosis and brief episodes could once be given the option of careful, supervised treatment discontinuation. The potential salutary effect of antidepressants during the psychotic prodrome and presence of characteristics differentiating patients at risk for schizophrenia or bipolar disorder should be investigated further.
Psychotic disorder not otherwise specified (PsyNOS) and brief psychotic disorder (BrPsy) are underresearched, mutually exclusive diagnostic entities that involve the presence of at least one positive psychotic symptom (i.e., non-bizarre delusions, hallucinations, disorganized speech, or grossly disorganized or catatonic behavior) that do not fulfill criteria for schizophrenia or schizoaffective disorder, a psychotic mood disorder, delusional disorder, acute psychotic disorder, schizophreniform disorder, or a psychotic disorder secondary to a medical disorder or substance use (American Psychiatric Association, 1994; Table 1). In addition, BrPsy requires a sudden onset of the psychotic symptom(s), a duration of psychosis lasting between 1 day and less than 1 month, and the eventual return to the premorbid level of functioning (American Psychiatric Association, 1994; Table 1). In clinical practice, the diagnosis of PsyNOS is often used when there is inadequate or insufficient information about the psychotic symptom(s) to make a specific diagnosis or when the psychotic symptoms do not fit any specific psychotic disorder.
While presenting a diagnostic problem due to their high rates of diagnostic instability ranging from 23% to 87% (Fennig et al. 1995; Susser et al. 1995; Jorgensen et al. 1996; Kumra et al. 1998; McClellan and McCurry, 1999; Schwartz et al. 2000; Nicolson et al. 2001; Pillmann et al. 2002; Saijith et al. 2002), these underresearched diagnostic entities provide an opportunity to study risk factors for the development of specific syndromal psychotic disorders, such as schizophrenia or bipolar disorder. In addition, despite conflicting data about the nature of the outcome of subjects with PsyNOS or BrPsy and the reliability of progression to a schizophrenia-spectrum disorder, several studies seem to suggest a dimensional symptomatic and genetic closeness to schizophrenia (Kendler and Walsh 1995; Kumra et al. 1998; Kumra et al. 2000; Kumra et al. 2001). On the other hand, however, two recent studies have also reported significant overlap of biological markers of both psychosis NOS and schizophrenia with bipolar disorder, including neurocognitive deficits in memory and attention (McClellan et al. 2004) and polymorphisms in the 13q33.2 gene G72/G30 (Addington et al. 2004). Therefore, the study of patients with PsyNOS and BrPsy provides an opportunity to study risk markers that predict future progression to schizophrenia as opposed to bipolar disorder.
In this context, it is surprising that, except for the Zucker Hillside Hospital Recognition and Prevention (RAP) Program (Correll et al. 2005), PsyNOS and BrPsy have not been studied in their own right within the context of prodromal schizophrenia research. Rather, in prodromal schizophrenia research, subjects exhibiting psychotic symptoms of self-limited duration of either 6 days (Yung et al. 1998; McGorry et al. 2003) or 3 days (McGlashan et al. 2003; Woods et al. 2003) per week are mixed in with patients exhibiting attenuated symptoms or a recent deterioration in mental state or functioning in the presence of genetic risk. Furthermore, subjects with psychotic symptoms for more than 1 week, also fulfilling Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) (American Psychiatric Association 1994) criteria for PsyNOS or BrPsy, have also been used as a psychosis “outcome” group, clinically requiring interventions (McGorry et al. 2002). To date, there is no evidence supporting either classification system of individuals fulfilling criteria for PsyNOS or BrPsy. Therefore, clarity about the final outcome and potential predictors of subjects with psychosis not fulfilling criteria for schizophrenia, schizoaffective disorder, or a psychotic mood disorder has the potential to enhance the specificity of prodromal psychosis research.
Previous studies that followed patients with PsyNOS or BrPsy have focused on diagnostic stability and correlates of symptomatic outcome (Fenton and McGlashan 1989; Fennig et al. 1995; Susser et al. 1995; Jorgensen et al. 1996; Kumra et al. 1998; Susser et al. 1998; McClellan and McCurry 1999; Lewis et al. 2000; Schwartz et al. 2000; Nicolson et al. 2001; Pillmann et al. 2002; Saijith et al. 2002; Weiser et al. 2003; Hlastala and McClellan 2005). In these studies, the diagnostic stability of PsyNOS and BrPsy was relatively low, and similar outcome predictors than those known for schizophrenia were identified; i.e., higher baseline symptom ratings, premorbid motor impairments, drug use, and lower verbal as well as full-scale intelligence quotient (IQ) (Fenton and McGlashan 1989; Lewis et al. 2000; Nicolson et al. 2001; Weiser et al. 2003). However, none of the studies to date has attempted to identify predictors for conversion to specific syndromal psychotic disorders, such as schizophrenia or bipolar disorder, although this may help guide targeted preventive and symptomatic interventions in truly prodromal patients. Furthermore, none of the studies to date has examined the predictors for full symptomatic remission, including attenuated positive and negative symptoms. We have previously reported on the baseline criteria of 29 youths with a diagnosis of PsyNOS and BrPsy, as well as on the diagnostic stability and symptomatic as well as syndromal outcomes in the 26 youngsters with at least 6 months of follow up (Correll et al. 2005). In view of the variable outcomes observed in this study, the aim of this paper was to examine predictors of conversion to schizophrenia, bipolar disorder, and full symptomatic remission in the 26 youths with PsyNOS and BrPsy who were followed prospectively in a schizophrenia prevention program.
Individuals in this study represent a consecutively enrolled subsample drawn from a large prospective, naturalistic study, currently underway at the Zucker Hillside RAP Program. The RAP program focuses on the course and outcome of help-seeking adolescents considered to be prodromal for schizophrenia (Cornblatt et al. 1998; Cornblatt 2002; Cornblatt et al. 2002; Cornblatt et al. 2003; Lencz et al. 2003; Lencz et al. 2004). Patients for this report were recruited between January, 1998, and January, 2004, largely from mental health referrals to The Zucker Hillside Hospital/Schneider Children's. With the exception of targeted treatment trials, the pharmacologic and psychotherapeutic treatment followed a naturalistic, symptom-based framework that is derived from the patient's clinical need and best-practice guidelines. Treatment targeted positive and negative psychotic symptoms as well as co-morbid depression, anxiety, and attentional and behavioral disturbances. The research protocol was approved by the Institutional Review Board at the Long Island Jewish Medical Center. Written, informed consent was obtained from the patient if 18 years of age or older, or from the parent (with written assent from the patient) if under 18 years.
Subjects included in this study were youths ages 12–22 years with a DSM-IV diagnosis of PsyNOS or BrPsy defined by: (1) failure to meet A criteria for schizophrenia, i.e., presence of only one nonbizarre psychotic symptom; (2) failure to meet B criteria for schizophrenia, i.e., presence of sustained adequate role functioning; and/or (3) failure to meet C criteria for schizophrenia, i.e., presence of psychotic episodes lasting less than 1 month (in the absence of any psychotropic treatment). Subjects meeting the third criterion, reflecting short duration, all met criteria for BrPsy and were so labeled. Exclusion criteria were: (1) meeting criteria for schizophrenia, schizophreniform disorder, bipolar disorder with psychotic features, major depressive disorder with psychotic features, or psychotic disorder secondary to a medical condition or substance use; (2) history of neurological, neuroendocrine, or other medical disorder known to affect the brain; (3) IQ<70; and (4) severe or imminent risk of harm to self or others.
Demographic and background information was obtained from the parent/legal guardian, including past treatment history and socioeconomic status (SES), which was categorized according to Hollingshead (1957), ranging from 1 (highest) to 5 (lowest). Diagnostic interviews were first administered to the parent/legal guardian to provide information useful for probes when interviewing the patients and to alleviate patients' potential concerns about confidentiality. Following the informant's interview, the same rater interviewed the patient and sought to clarify any discrepancies with the parental information.
Diagnoses of DSM-IV Axis I and Axis II disorders were based on the Schedule for Affective Disorders and Schizophrenia for School-Age Children, Epidemiologic Version (K-SADS-E; Orvaschel and Puig-Antich 1994) and the Structured Interview for DSM-IV Personality (SIDP-IV) (Pfohl et al 1995). In addition, the presence and severity of positive and negative psychotic symptoms, including subthreshold, attenuated levels, were rated using the Scale of Prodromal Symptoms (SOPS), a research instrument designed specifically to assess attenuated schizophrenia-like symptomatology for identifying prodromal states (McGlashan et al. 2001; Miller et al. 2002). On the basis of SOPS ratings, patients enrolled into the RAP program were classified as belonging to one of the following three clinical high risk groups for schizophrenia: (1) Clinical high-risk group with attenuated negative symptoms (CHR−), defined by the presence of SOPS rated negative and nonspecific symptoms (scale of 0–6), including (i) social isolation/withdrawal; (ii) avolition; (iii) decreased expression of emotion; (iv) decreased experience of emotion and self; (v) decreased ideational richness; and (vi) deterioration in role functioning (i.e., at least one negative symptom SOPS item scored at 3 (i.e., moderate) or higher and absence of a rating of more than 2 (i.e., mild) on any positive symptom item) (Lencz et al. 2004). (2) Clinical high-risk, attenuated, positive symptom groups (CHR+), defined by a rating of least a 3 (i.e., moderate) on one or more SOPS-positive symptoms (scale of 0 [not present] to 6 [extreme or psychotic intensity, which is exclusionary for CHR+, even if present only briefly]), including: (i) unusual thought content; (ii) suspiciousness; (iii) grandiosity; (iv) perceptual abnormalities; and (v) conceptual disorganization. (3) Schizophrenia-like psychosis (SLP), defined by positive symptoms of psychotic intensity (i.e., SOPS-positive symptom rating of 6) that are either not severe enough (i.e., only one nonbizarre psychotic symptom and/or absence of clinically relevant social, academic, or vocational impairment, despite being “severe and psychotic” and being associated with “usual” or “likely” interference with thinking, beliefs, feelings, social relations, behavior, and function), or that are present too briefly (i.e., less than 1 month) to fulfill criteria for schizophrenia. All patients described in this paper belonged to the latter group at entry into the RAP program, meeting criteria for BrPsy (in case of rapid onset and eventual full return to baseline functioning) or PsyNOS (all other subjects not fulfilling criteria for schizophrenia or another major DSM-IV psychotic disorder).
SOPS ratings and the classification of the patients in this study were based on information elicited by the companion interview, the Structured Interview for Prodromal Syndromes (SIPS) (McGlashan et al. 2001; Miller et al. 2002) (n=17), or on the basis of all other available information for those subjects entering the study before the development of the SIPS (n=9). For subjects indicating psychotic-level positive symptomatology on the SOPS, the K-SADS-E was used to delineate DSM-IV criteria for a syndromal psychotic disorder. All SOPS and diagnostic ratings were made on the basis of all available information by a RAP consensus conference. All semistructured interviews and cognitive testing were administered by trained masters- or doctoral-level psychologists. As reported by Lencz et al. (2004), a high interrater reliability was established for all individual SOPS items (interclass correlation coefficients>0.80 for most items and at least 0.67 for all items) and 100% concordance for diagnostic stage.
In addition to prodromal and syndromal psychotic symptoms, depressive symptomatology was assessed with two self-report measures, the Children's Depression Inventory (CDI; Kovacs, 1985) for children up until age 16, and the Beck Depression Inventory (BDI; Beck et al., 1988a) for youths age 17 and above. The CDI is a 27-item, self-report rating scale for school-aged children and the BDI is a 21-item, self-report rating scale designed for older adolescents and adults. Endorsed symptoms of depression were scored from 0 to 3 (BDI) and 0 to 2 (CDI), with higher scores representing more severe symptoms. Anxiety symptoms were rated using the Beck Anxiety Inventory (BAI; Beck et al., 1988b), a 21-item, self-report scale that has scores ranging from 0 (not at all) to 3 (severe). Patients also underwent IQ testing, using the Wechsler Adult Intelligence Scale-Revised (WAIS-R; Wechsler 1981) for youngsters ages 16 and older and the Wechsler Intelligence Scale for Children-Third Edition (WISC-III; Wechsler 1991) for subjects age 12–15 years. Full-scale IQ was estimated with a two-subtest short form using Vocabulary and Block Design Subtests. These two subtests, individually, have excellent reliability and correlate highly with the full-scale IQ over a wide age range. The IQ estimate derived from these scores has satisfactory reliability (r=0.91) and validity (r=0.86) (Sattler 2001).
Finally, patients were administered an extensive neurocognitive test battery. For the purpose of this paper, we a priori selected only three cognitive tests, the Continuous Performance Test–Identical Pairs Version (CPT-IP) (see Cornblatt et al. 1988; Cornblatt et al. 1989; Cornblatt and Kelip 1994; Cornblatt and Malhotra 2001), the Wisconsin Card Sorting Test (WCST; Heaton and Pendleton 1981; Goldberg and Weinberger 1988; Heaton et al. 1999), and the California Verbal Learning Test (CVLT; Delis et al. 1987). CPT-IP version is a cognitively challenging version of the traditional CPT that measures sustained visual attention The CPT-IP requires a subject to respond whenever two identical stimuli appear in a row within a sequence of rapidly flashed trials. Four subtests with stimuli (2 digits, 3 digits, 4 digits, and shapes) assessing increasing processing load were examined. The primary measure of interest is d-prime—a signal detection theory-derived index of discriminability of target from non-target trials. The WCST is a widely used measure of executive function (Grant 1948; Heaton et al. 1999), which measures the ability to generate hypotheses, establish response sets, and switch sets by sorting stimulus cards based on three attributes (color, form, and number). Testing is completed when the participant sorts six categories (based on 10 consecutive correct responses) or reaches a maximum number of 128 trials. The number of perseverative errors is the dependent measure here. The CVLT is a neuropsychological test to assess verbal learning and memory. Participants recall lists of words over five repeated trials and over short and long delays. Word lists consist of 16 common words belonging to four categories. The number of words recalled over the five repeated trials is the dependent variable here.
We extracted these three cognitive measures for this study, as the CPT (Levy et al. 1993; Cornblatt and Kelip 1994; Cornblatt and Obuchowski 1997), WCST (Mohamed et al. 1999; Bilder et al. 2000), and CVLT (Bilder et al. 2000; Fitzgerald et al. 2004) are among the most established measures of cognitive abnormalities in schizophrenia. Moreover, in a previous analysis of patients with attenuated psychotic symptoms, we found that attentional abnormalities measured by the CPT did not predict conversion to psychosis, but they were a required underlying trait, whereas abnormalities in executive functioning appeared to be associated with the emergence of severe positive symptoms (Cornblatt et al. 2003). Furthermore, verbal memory (measured largely by the CVLT) was the sole cognitive variable that predicted conversion to psychosis in the RAP Program sample putatively prodromal for schizophrenia, consisting of subjects with attenuated positive symptoms (Lencz et al. 2006)
Follow-up ratings were made approximately 6 months after entry into the RAP Program, regularly every 6–9 months, and at termination of treatment or conversion to a syndromal psychotic disorder. Although the follow-up periods of 6–9 months reduce the accuracy of recall of psychotic symptoms, such intervals represent a pragmatic standard in prospective studies of this kind. The diagnostic outcome of each subject with follow-up data of at least 6 months (unless conversion to a syndromal psychotic disorder occurred) was examined with respect to SOPS item anchors, representing the level of prodromal and syndromal psychosis, as well as DSM-IV criteria for major psychotic disorders. An individual SOPS-positive symptom score of 6 (i.e., severe) represented presence of psychosis. A rating of 3–5 (mild, moderate, or moderately severe) on the SOPS-positive symptom score represented presence of attenuated positive symptoms. A rating of 3–5 on the SOPS-negative symptom score represented presence of attenuated negative symptoms. A score of 2 or below (i.e., mild, minimal/questionable, or absent) on both the SOPS-positive and the SOPS-negative symptom score represented full remission from all prodromal psychotic symptoms. Previous analyses showed a 100% concordance of the consensus ratings for each prodromal and psychotic symptom outcome stage (Lencz et al. 2004). Follow-up consensus ratings in the 26 patients were made unblinded to baseline diagnoses and, on the basis of all available clinical information from clinician reports, telephone interviews, and in-person follow-up interviews of patients and/or their caregivers. As part of the routine RAP procedures, all prodromal and DSM-IV baseline and follow-up diagnoses were subsequently validated by independent, consensus diagnosticians blinded to research assigned diagnoses.
Distribution of baseline data was analyzed using descriptive statistics. Between-group comparisons of continuous variables were performed using analyses of variance (ANOVA). Between-group comparisons of dichotomized or categorical variables were performed using either chisquared statistics or the Fischer exact test, whenever individual numbers per cell were too small to allow using chisquared statistics. To ensure comparability of the CDI and BDI, the two age-dependent measures for depression across the entire sample, a total depression score was calculated as a percentage of the maximum possible score on each of these scales. To allow easy comparison of mean levels of anxiety with the mean level of depression, a total anxiety score was also calculated as a percentage of the maximum possible score on the BAI. Cognitive performance indices were converted into z-scores, based on published norms for the WCST and CVLT and a normative database for the CPT-IP. All analyses were two-sided, with α= 0.05. Post hoc comparisons of baseline and treatment predictors between individual outcome groups were conducted whenever the overall comparison between the four main outcome groups was significant at a level of p <0.2. This more liberal threshold was used due to the small sample size and exploratory nature of the analyses. Furthermore, we conducted correlelational analyses between baseline, treatment, and outcome variables and duration of follow up, because this is a potential confounding variable of other associations. In addition, multiple predictor variables for one outcome group were tested for intercorrelations. Because the analyses were considered preliminary and hypothesis generating, we did not control for multiple comparisons. Data were analyzed using JMP 5.0.1, 1989–2003 (SAS Institute Inc.).
Twenty six subjects with a diagnosis of PsyNOS (n=22) or BrPsy (n=4) were prospectively studied for at least 6 (mean 22.8±19.4) months. The mean age of the sample was 15.9±2.7 (range 12–22) years; 65.4% were male and 50.0% were Caucasian; and the mean IQ was 97.1±20.3 (Table 2). The main co-morbid conditions (mean number 2.1±2.4) included depressive disorders (53.8%), fulfilling DSM-IV criteria for personality disorders (42.3%), anxiety disorders (38.5%), oppositional defiant and conduct disorder (34.6%), and attention-deficit/hyperactivity disorder (30.8%).
Despite the frequency of lifetime mood and anxiety disorder, baseline self-ratings of depression and anxiety were low with a mean of 21.3±17.7% of the maximum score on the BDI or CDI, and of 12.5±9.9% of the maximum score on the BAI, respectively. Consistent with presence of suprathreshold psychotic symptoms, the mean total SOPS-positive symptom score of 14.2±4.5 was relatively high (maximum score, 25). The entire group of patients with PsyNOS or BrPsy was about 1 standard deviation (SD) below an aged-matched norm for attention (CPT-IP z-score,−0.95±1.1) and verbal learning (CVLT z-score,−0.97±1.9, and slightly above the aged-matched norm for executive functioning (WCST z-score, 0.13±1.30), with large variability between the outcome groups.
At follow up, four main outcome groups emerged from the data: (1) progression to schizophrenia or schizoaffective disorder (n=7, 27.0%); (2) progression to bipolar disorder (n=4, 15.4%), either with (n=3, 11.5%) or without (n=1, 3.8%) psychotic features; (3) residual attenuated positive or negative symptoms (n=11, 42.3%); and (4) full remission of all attenuated psychotic symptoms (n=5, 19.4%). Significant differences between the four groups were present for Hispanic ethnicity (p=0.0089), diagnosis of either PsyNOS or BrPsy (p=0.017), and presence of anxiety disorders (p=0.013) (Table 2).
In the naturalistic setting of this study, 88.5% of patients with PsyNOS or BrPsy received psychotropic drugs for a mean of 14.1±13.6 months (Table 3). The remaining 11.5% refused psychotropic treatment. Most patients received atypical antipsychotics (n=20, 76.9%), followed by antidepressants (n=15, 57.7%). Treatment with anxiolytics/hypnotics (n=4, 15.4%), psychostimulants (n=4, 15.4%), and/or mood stabilizers (n=3, 11.5%) was less common. Nonadherence to psychotropic drug treatment, defined as stopping the medication for at least 2 weeks, occurred in 56.5% of the 23 patients with available data. Seventeen of 25 patients with available information (68.0%) were off all psychotropic medications at the last follow-up appointment, with 19 (76.0%) being off antipsychotics. Although there were numeric differences in the treatment status at end point and the duration of psychotropic drug treatment across the different outcome groups, with mostly shorter durations in the schizophrenia/schizoaffective group and longer treatment in the bipolar disorder and full remission group, none of these differences reached statistical significance.
Except for a correlation between higher SOPS total negative symptom scores and shorter follow-up duration (p = 0.012), none of the baseline, treatment, and outcome variables was significantly associated with follow-up duration. Table 4 summarizes the significant predictors of three main outcome groups: (1) progression to a syndromal psychotic disorder (n = 10, 38.5%) (i.e., schizophrenia (n=6, 23.1%), schizoaffective disorder (n=1, 3.8%), or psychotic bipolar disorder (n=3, 11.5%); (2) schizophrenia or schizoaffective disorder (n=7, 26.9%); (3) bipolar disorder (n=4, 15.4%); and (4) full remission of all suprathreshold and attenuated positive and/or negative psychotic symptoms (n=5, 19.4%) at follow up. Because no significant predictors were found for remission of suprathreshold psychotic symptoms with continued presence of residual attenuated positive and/or negative symptoms (n=11, 42.3%), this group is not represented in Table 4.
Progression to a syndromal psychotic disorder was predicted by fulfilling DSM-IV criteria for schizotypal personality disorder at baseline (30.0% vs. 0.0%, p=0.046, Fischer exact test). Development of schizophrenia/schizoaffective disorder was associated with absence of anxiety disorders (0.0% vs. 38.5%, p=0.023, Fischer exact test) and worse executive functioning measured via perseverative error rate on the Wisconsin Card Sort Test and expressed as z-scores at baseline compared to the rest of the sample (−1.12±1.46 vs. 0.52±1.33, F(1): 5.547, p= 0.029) and compared to patients who developed any bipolar disorder−1.12±1.46 vs. 1.10±1.02, F(1): 6.57, p=0.037). Absence of anxiety disorders was not associated with lower executive functioning as measured by the Wisconsin Card Sort Test compared to presence of an anxiety disorder (−0.07±1.7 vs. 0.46±1.1, p= 0.45).
In contrast, conversion to a psychotic or nonpsychotic variant of bipolar disorder was predicted by presence of anxiety disorders. This was true compared to the rest of the sample (100.0% vs. 27.3%, p= 0.014, Fischer exact test) and compared to the subgroup of patients who developed schizophrenia or schizoaffective disorder (100.0% vs. 0.0%, p=0.0030, Fischer exact test). To examine whether antidepressant use associated with a diagnosis of anxiety disorders, rather than anxiety disorder itself, was responsible for this result, we examined the relationship between anxiety disorder diagnosis and antidepressant use. In patients with and without anxiety disorders, neither the frequency 70.0% vs. 50.0%, p=0.32) nor the duration (4.9±6.7 months vs. 5.6±6.57 months, p=0.82) of antidepressant use was significantly different.
Remission of any psychotic symptoms, even in attenuated form, was predicted by Hispanic ethnicity (60.0% vs. 4.8%, p=0.0047, Fischer exact test); an initial diagnosis of BrPsy (80.0% vs. 9.5%, p= 0.014, Fischer exact test), longer antidepressant treatment (11.2±9.0 months vs. 3.8±5.2 months, F(1)=5.11, p=0.035), and better attention at baseline as measured by the 2-digit span continuous performance test (0.20±1.2 vs.−1.2±1.2, F(1)=4.69, p= 0.042). Clinical outcome of any category was not significantly associated with any of the psychotropic medication treatment variables, including medication status at baseline or end point, duration of psychotropic drug treatment other than antidepressants, and medication nonadherence status. Furthermore, except for the association between Hispanic ethnicity and a diagnosis of brief psychotic disorder (r2=0.22), none of the four predictor variables correlated with each other to a relevant degree,; i.e., BrPsy vs. antidepressant treatment duration, r2=0.0092 (with numerically shorter (not longer) antidepressant treatment duration in the BrPsy group); Hispanic vs. attention, r2=0.019; BrPsy vs. attention, r2=0.094; Hispanic vs. antidepressant treatment duration, r2=0.014; and attention vs. antidepressant treatment duration, r2=0.000001.
Finally, antidepressant use was present in 3 of the 4 patients who ultimately developed bipolar disorder. In 2 of these 3 patients, antidepressant-related hypomania was observed within 2 and 6 weeks of antidepressant treatment, respectively. In the first case, the antidepressant was stopped 5 weeks before the first manic episode. In the second case, the antidepressant was switched and low-dose antipsychotic treatment (quetiapine 50mg/day) was started, but the patient developed full mania 10 weeks later on this treatment regimen. By contrast, hypomania or activation syndromes were absent in all other 12 patients in this sample with PsyNOS and BrPsy who received treatment with antidepressants and who did not develop a bipolar spectrum disorder.
This preliminary study of outcome predictors in adolescents and young adults with PsyNOS and BrPsy followed prospectively in a schizophrenia prodromal research program resulted in several findings: (1) Progression to a syndromal psychotic disorder (i.e., schizophrenia, schizoaffective disorder, or bipolar disorder with psychotic features) was associated with fulfilling DSM-IV criteria for a schizotypal personality disorder at baseline; (2) conversion to schizophrenia or schizoaffective disorder was associated with the absence of anxiety disorders and poor executive functioning, measured by the preservative errors on the WCST; (3) progression to bipolar disorder with or without psychotic features was associated with presence of anxiety disorders; and (4) recovery from any positive and negative psychotic symptomatology was predicted by Hispanic ethnicity, an initial diagnosis of BrPsy, longer duration of antidepressant treatment, and better baseline attention as measured by the 2-digit span Continuous Performance Test. These results extend the findings from our previous report (Correll et al. 2005), following up on the initial finding that BrPsy was associated with a greater chance of full and attenuated psychotic symptom recovery. Moreover, this study suggests several additional potential markers of diagnostic and symptomatic outcome in patients initially diagnosed with PsyNOS or BrPsy.
Fulfilling DSM-IV criteria schizophrenia-spectrum personality disorders has previously been reported as a risk factor for the development of schizophrenia (Fenton and McGlashan 1989; Siever et al. 1990; Chapman et al. 1994; Kwapil et al. 1997; Walker et al. 1999; Weiser et al. 2002; Neumann and Walker 2003; Siever and Davis 2004; McGlashan et al. 2005). In this context, our finding reinforces the idea of a psychosis-prone spectrum that includes cluster A personality disorders and validates currently used selection criteria that include schizotypal personality as a risk group in some prodromal schizophrenia clinics (Seeber and Cadenhead 2005). However, an overlapping risk for bipolar disorder has also been described for patients diagnosed with schizotypal personality disorder (Kety et al. 1994; Lyons et al. 1994; Erlenmeyer-Kimmling 2000). Thus, our finding that schizotypal personality disorder features also increased the risk for psychotic bipolar disorder highlights the potential overlap between syndromal (Addington et al. 2004; McClellan et al. 2004) and prodromal (Thompson et al. 2003; Correll et al. 2007a) symptoms in schizophrenia and bipolar disorder.
The association between poor executive functioning, assessed with the WCST, and progression to schizophrenia or schizoaffective disorder is consistent with extensive literature regarding poor executive functioning and schizophrenia (Heaton and Pendleton 1981; Goldberg and Weinberger 1988; Cannon et al. 2000; Bilder et al. 2000; Sitskoorn et al. 2004; Hughes et al. 2005), although findings in their unaffected family members have been less clear (Stratta et al. 1997; Laurent et al. 2001; Wolf et al. 2002). However, this is the first report to suggest that lower executive functioning could be used as an additional risk marker to identify subjects at ultrahigh risk for schizophrenia, identified via presence of psychotic symptoms not fulfilling criteria for schizophrenia or schizoaffective disorder. However, whether presence or degree of executive functioning deficits can serve as a specific risk marker for the conversion to schizophrenia, or whether it may signal a risk for a psychotic disorder in general, requires further study, because impaired executive functioning has also been described in bipolar disorder (Quraishi and Frangou 2002; Altshuler et al. 2004). The significant association between the lack of co-morbid anxiety disorders and risk for schizophrenia compared to the rest of the subjects with PsyNOS or BrPsy is not fully consistent with the existing literature, because schizophrenia has also been associated with co-morbid anxiety disorders (Cosoff and Hafner 1998; Braga et al. 2004; Pallanti et al. 2004). Therefore, it is unclear why none of the youngsters in our sample who ultimately progressed to schizophrenia or schizoaffective disorder had a lifetime diagnosis of any anxiety disorder. This may be due to our relatively young patient population, some of whom may develop anxiety disorders after the progression to schizophrenia, or it may be due to a cohort effect in our small study sample.
Thus, the finding of a lack of anxiety disorders predicting schizophrenia development in our sample may not generalize to the heterogeneous group of patients with schizophrenia and could be due to an ascertainment bias toward inclusion of more insidious onset cases with predominant negative symptoms. Furthermore, the lack of association between verbal learning memory and progression to schizophrenia is discordant with previous findings from high-risk groups (Whyte et al. 2006; Brewer et al. 2006; Lencz et al. 2006; Niendam et al. 2006). However, this association has not always been robust (Whyte et al. 2006) and has been demonstrated in patients with attenuated positive symptoms. Because patients with schizophrenia have marked deficits in verbal learning and memory (Bilder et al. 2000; Fitzgerald et al. 2004), it is possible that the presence of psychosis, albeit at a level below the threshold for schizophrenia, interfered with verbal learning at baseline to a degree that obscured differences between outcome groups. This is suggested by the impairments on the CVLT across all groups that ranged from−0.3 to−1.6 z-scores, with an overall group impairment of 1 SD.
On the other hand, the suggested increased risk for developing bipolar disorder in the presence of a lifetime diagnosis of anxiety disorders is consistent with a large body of studies showing high rates of co-morbid anxiety disorders in patients with bipolar disorder (Tamam and Ozpoyraz 2002; Sasson et al. 2003; Boylan et al. 2004; Simon et al. 2004). This association seems to be particularly strong in patients with early- and very early-onset bipolar disorder (Birmaher et al. 2002; Wozniak et al. 2002; Mick et al. 2003; Perlis et al. 2004). However, the severity of self-rated anxiety symptoms on the Beck Anxiety Inventory was not significantly different between patients who developed schizophrenia or bipolar disorder and those who did not. This lack of association with symptom severity could be due to severity fluctuations, the confounding of treatments received at the time of the ratings, or the lack of sensitivity of self-rating scales in psychotic youths. Nevertheless, the presence of a lifetime anxiety disorder diagnosis in all subjects who developed bipolar disorder may represent a greater affective responsivity that could increase the risk for the development of bipolar disorder in subjects presenting with symptoms that are consistent with the schizophrenia prodrome.
It is important to note, though, that patients who later progressed to bipolar disorder did not differ from those developing schizophrenia in any of the baseline variables, except for a higher prevalence of anxiety disorders. This inability to distinguish patients destined to develop bipolar disorder from those progressing to schizophrenia-spectrum disorders has been described in a small case series from another prodromal schizophrenia study (Thompson et al. 2003). Although this initial similarity in the symptom presentation could be limited to a subgroup of patients who are recruited into programs that select patients based on the theoretical risk for schizophrenia, the prospective assessment of the bipolar prodrome and its relationship to anxiety disorders deserves further study to allow for more targeted early preventive interventions (Correll et al. 2007a,b). In particular, other putative risk markers described previously should be investigated, including psychiatric family history, presence and severity of mania-like symptoms, temperamental characteristics, affect recognition and regulation, circadian rhythm, anatomical and neurochemical brain abnormalities (Correll et al. 2007b).
Full and prodromal psychotic symptom remission was predicted by Hispanic ethnicity, better attention as measured by the 2-digit CPT, longer duration of antidepressant treatment, and a baseline diagnosis of BrPsy. Our finding of a significant effect of Hispanic ethnicity on remission from any suprathreshold as well as attenuated psychotic symptoms could be due to several factors, including chance due to the relatively small sample size, diagnostic bias, or differences in symptom expression in Hispanics. Several studies suggest that compared to other ethnic groups, Hispanics may present with more nonspecific diagnoses (Matthews et al. 2002) and are diagnosed with less psychotic and more affective disorders (Flaskerud and Hu 1992), despite the fact that Hispanics had significantly higher levels of self-reported psychotic symptoms (Minsky et al. 2003). The latter may either indicate a diagnostic bias or culturally determined expression of psychotic symptoms with less prognostic stability, which can create the problem of inaccurate classification and interpretation of psychotic symptoms among Hispanics, as discussed in detail in the appendix of the DSM-IV (American Psychiatric Association 1994), which focuses on cultural factors affecting diagnostic evaluation. In this context, the clinical picture of “ataque de nervios” may also be a relevant and culturally bound syndrome (López and Guarnaccia 2000) that should be considered in clinical high-risk schizophrenia studies.
The association between better attention and symptom remission mirrors the prevalence of attentional deficits measured by CPT-IP in patients with chronic schizophrenia (Cornblatt et al. 1989; Cornblatt et al. 1997; Chen et al. 2004), adolescents and adults in the earliest stages of illness (Cornblatt et al. 1997; Cornblatt et al. 1998), nonpsychotic relatives of schizophrenia patients (Franke et al. 1994; Laurent et al. 1999; Chen et al. 2004), and individuals with schizotypal personality disorders (Lenzenweger et al. 1991; Roitman et al. 1997).
The association between longer antidepressant treatment and full symptomatic remission requires further attention. Recently, we have shown similar results in a group of naturalistically treated patients from our larger sample of youth at clinical risk for schizophrenias who were identified by presence of attenuated positive symptoms and absence of suprathreshold psychosis (Cornblatt et al. 2007). This is of relevance because antipsychotics, used predominantly in prodromal schizophrenia intervention studies (Correll and Kane 2004), are associated with weight gain and metabolic abnormalities that seem to be accentuated early on in life and in the treatment history (Correll and Carlson 2006; McGlashan et al. 2006). Although it is possible that the nonrandom treatment assignment could have led to a differentially higher antidepressant treatment of patients with less risk for the development of psychosis, it is also possible that antidepressants exert some neuroprotective action and/or reduce stress that can result in a reduction of progression to full psychosis in the early disease stages (Li et al. 2000; Michael-Titus et al. 2000; Dazzi et al. 2001). This possibility requires further study in prodromal psychosis research settings.
On the other hand, antidepressant treatment could also be a marker of some other “protective” factors, such as presence of mood symptoms, less negative symptoms, less clinical severity, or greater treatment adherence. Furthermore, 3 of the 4 patients who developed bipolar disorder also received antidepressant treatment, and 2 of those developed antidepressant-induced hypomania prior to the development of their first episode of mania. Thus, in patients with a diathesis for bipolar disorder, antidepressant use may uncover the underlying risk for bipolar disorder, and aggressive mood-stabilizing treatment should be initiated in these cases. This underscores the need for research aiming to identify variables that can help to distinguish patients at risk for bipolar disorder from those at risk for schizophrenia (Correll et al. 2007a).
Finally, our finding that remission of all psychotic and prodromal symptoms in 3 of the 4 patients with a diagnosis or BrPsy is in line with previous reports of favorable outcome in patients with BrPsy (Fennig et al. 1995; Susser et al. 1995; Jorgensen et al. 1996; Susser et al. 1998; Pillmann et al. 2002). If reconfirmed in larger samples, this has implications for selection criteria used to enrich samples at risk for schizophrenia, as it challenges the practice of categorizing patients with brief psychotic episodes as being prodromal for schizophrenia. Conversely, our findings also challenge the opposite practice of categorizing individuals with psychotic symptoms lasting 1 week or more as having reached a psychotic disorder end point (McGorry et al. 2002). Although treatment is indicated at that time, patients should be followed beyond that point to investigate whether, in fact, subjects are progressing to schizophrenia and, if so, whether or not the course of the illness can be influenced by means of early interventions. Although treatment may be seen as a relevant confound in this observation, in our sample, pharmacologic treatment at baseline or follow up was not significantly associated with any of the categorical outcomes. Nevertheless, even though we observed a relatively high nonadherence rate in our sample (56.5%), which may have allowed us to observe the natural course of these disorders over time, we cannot fully exclude that this negative finding was due to limited power in our study or to a potential carry over effect of treatment received at some point during the follow-up duration.
It is important to note that the presented findings are preliminary and need to be interpreted with caution. Limitations include the small number of patients, both in the entire sample and in specific outcome groups, lack of a matched control group, different durations of naturalistic follow up, uncontrolled medication effects, lack of specific assessments of negative symptoms, irritability, and psychosocial, educational, and vocational functioning, as well as the sample ascertainment from one single center, which may reduce the generalizability of the results. Furthermore, that age was not significantly related to outcome may have been due to the fact that we studied a narrow and predominantly adolescent age group. Although results need to be followed up by future research, this is the first study that followed patients with PsyNOS and BrPsy as separate risk groups in the context of schizophrenia prodrome research. It is also the first study that examined demographic, clinical and limited neuropsychological predictors of diagnostic and symptomatic outcomes in patients with these two under-researched diagnostic entities.
This preliminary study of youths with PsyNOS or BrPsy followed in a prodromal schizophrenia research setting for a mean of more than 2 years confirms prior research regarding variable outcomes, including progression to both affective and nonaffective psychotic disorders in a minority of patients. The fact that progression to schizophrenia/schizoaffective disorder or psychotic bipolar disorder was associated with presence of schizotypal personality disorder criteria indicates that in some patients certain cluster A personality features presenting early in life may represent a risk state for a syndromal psychotic disorder, rather than being a diagnostic end stage. Thus, patients with schizotypal personality disorder features in adolescence should be studied as a potential risk group in studies assessing the risk for schizophrenia and bipolar disorder (Seeber and Cadenhead 2005). The finding that patients progressing to bipolar disorder did not differ from those developing schizophrenia, except for a higher lifetime prevalence of anxiety disorders, requires further investigation. Thus, these data suggest that a careful assessment of syndromal or subsyndromal anxiety disorders and symptoms should be undertaken in psychosis prodromal programs that should also broaden their focus to include the study of prodromal presentations and predictors for bipolar disorder (Correll et al. 2007b). Remission of any psychotic symptoms, even in attenuated form, was associated with Hispanic ethnicity, better attention at baseline, longer treatment with antidepressants, as well as acute onset and brief duration of psychosis, reflected by a diagnosis of BrPsy.
These results highlight the need for cultural awareness when diagnosing major mental disorders and studying prodromal illness manifestations. Moreover, after achieving full and sustained remission, patients with a sudden onset of psychosis of brief duration should be assessed to decide whether they could be offered once the option of careful and supervised treatment discontinuation. Future studies should examine potential interactions of psychotropic drug treatment and baseline predictions for conversion to bipolar disorder and other diagnostic and symptomatic outcomes in patients considered at risk for psychosis and/or bipolar disorder. In particular, research is needed to investigate the potential salutary effect of antidepressants during the psychotic prodrome and in patients with psychosis subthreshold for schizophrenia early on in life. In particular, questions regarding the need for continued antidepressant treatment beyond remission and regarding the mechanisms of such a potential effect should be pursued.
On the other hand, the potential risk of antidepressant treatment for uncovering a bipolar diathesis should also be considered. Finally, larger, long-term studies of individuals with PsyNOS, BrPsy, and additional populations considered at high-risk for the development of schizophrenia or bipolar disorder are needed to clarify and extend our preliminary findings of demographic, clinical, and neuropsychological predictors for symptomatic remission as well as the development of schizophrenia and bipolar disorder.
Dr. Correll has received speaker and/or consultancy honoraria from AstraZeneca, Bristol-Meyers Squibb, Eli Lilly, Intra-Cellular Therapeutics, Janssen, Organon, Otsuka, Pfizer, Solvay, Supernus, and Vanda. Dr. Shah has served on the speaker's Bureau of BMS/Otsuka, Janssen, and Shire. Dr. Kane has received speaker and/or consultancy honoraria from Abbott, AstraZeneca, Bristol-Myers Squibb, Eli Lilly, Janssen, Johnson & Johnson PRD, Otsuka, Pfizer, Inc., Wyeth, Lundbeck, Vanda, Astra-Zeneca, and PGxHealth. Dr. Cornblatt has served as a consultant to Janssen and Solvay and has received research support from BMS and Janssen. Dr. Auther has served as a consultant to Solvay. Dr. Smith, Mrs. McLaughlin, Dr. Foley, Ms. Olsen, and Dr. Lencz have no conflicts of interest or financial ties to disclose.
This work was supported in part by grant MH 61523-08 by the National Institute of Mental Health (NIMH) to Dr. Cornblatt, grant MH 65580 by the NIMH to Dr. Lencz, The Feinstein Institute for Medical Research, and The Zucker Hillside Hospital NIMH Advanced Center for Intervention and Services Research for the Study of Schizophrenia MH 074543-01 (Dr. Kane).