Early onset first episode psychosis: dimensional structure of symptoms, clinical subtypes and related neurodevelopmental markers

Early onset schizophrenia is a rare and severe condition in clinical and biological continuous with adult onset form of the disorder [1, 2]. The onset of schizophrenia in childhood is exceedingly rare (point prevalence <1/10.000 before the age of 12), but the incidence rises considerably in adolescence, and its prevalence is estimated at 0.23% in the age between 13 and 18 years [3, 4].

Studies exploring the effects of age of onset on clinical features in schizophrenia spectrum disorders showed that early and adult onset patients may differ in premorbid traits, illness presentation, clinical severity, and prognosis [3, 5‐7]. Nevertheless, the diagnosis of schizophrenia spectrum disorders with onset in children and adolescents, up to date, is made using the same DSM-5 criteria than in adults, as reported by the American Academy of Child and Adolescent Psychiatry, AACAP [8]. The main changes from DSM-IV-TR to DSM-5 in the chapter “Schizophrenia Spectrum and Other Psychotic Disorders” introduced a conceptual psychosis continuum emphasizing the dimensions of psychosis [9, 10], but still defines psychotic disorder categories and does not make allowance for potential characteristics of patients in childhood and adolescence [11].

The dimensional model proposes that symptoms of psychosis tend to cluster together within different symptom complexes which can co-exist in individual patients [12]; thus, a dimensional assessment can allow for a more specific and individualized clinical evaluation of patients with schizophrenia spectrum disorders [9]. This can be very helpful at the onset of psychotic disorders, considering that the heterogeneity of clinical presentation can make more complicated treatment decisions and different aspects of research [13]. A major issue within the dimensional approach is the careful identification of the psychopathological dimensions, so that their specific etiological, physiopathological and clinical correlates may be identified [12]. Many authors have conducted dimensional studies using the Positive and Negative Syndrome Scale (PANSS) [14] in adult populations of patients with schizophrenia spectrum disorders [15‐21]. The majority of dimensional studies agrees that 4 or 5 dimensions describe the psychosis construct, with positive, negative, disorganization and affective symptoms most frequently reported [22]. An important factor, which can influence the heterogeneity of results across studies, is the time frame of the study; a first episode psychosis (FEP) population may differ from a chronically ill population for symptoms profile and severity illness, so as the dimensional structure of FEP may be affected by early or adult age of onset [17, 22].

To our knowledge, very few studies have examined the dimensional structure of symptoms in childhood and adolescence psychosis, using different rating scales (Scale for Assessment of Positive Symptoms, Scale for Assessment of Negative Symptoms, Brief Psychiatric Rating Scale, Schedule for Affective Disorders and Schizophrenia-for School Age Children, PANSS) and includung patients with differnt diagnosis at baseline (schizophrenia and related disorders, mood disorders, anxiety disorders) [23‐26]. Only one among these examined the dimensional structure of symptoms in early onset psychosis through a principal component analysis (PCA) of the PANSS [26].

In the present paper, we studied a group of patients with FEP below 18 years, applying a three-step dimensional approach:

The main finding of this study was a four-factor model describing latent psychopathological dimensions resulting by the use of the PANSS in a sample of FEP patients with onset before 18. This solution can be considered an adequate model since it accounts for 47.15% of the total variance and the internal consistency for each factor is good. The four factors were “negative symptoms”, “delusions”, “conceptual disorganization” and “paranoid/hostility”, and each of them proved to identify a discrete clinical subgroup of patients by a cluster analysis. Furthermore, we found that delayed speech/language development was significantly associated with the “delusions” subtype, and that “conceptual disorganization” and “delusions” subtypes showed both a lower IQ than the other subtypes.

Factor analytic solutions of rating scale are in part scale dependent, disease dependent and state dependent; thus, variables as age of onset, diagnoses, acute or chronic phase of disease could modify the results between different studies [34]. Rapado-Castro et al. examined the dimensional structure of symptoms in early onset psychosis applying a PCA of the PANSS; they found a five-factor solution (positive, negative, depression, cognitive, hostility) suggesting a continuum between early and adult-onset form of the disorders [26]. Examining the four factors solution emerged from our study, we have found a substantial overlap with the Rapado-Castro model. “Negative symptoms” were the most consistent factor we found, accounting for the highest percentage (14.79%) of the total variance, as also reported by Rapado-Castro et al. Moreover, the negative dimension was found to be the most robust even in those studies exploring the dimensional structure of symptoms by the PANSS in samples of patients with adult onset first episode psychosis [16, 17, 35]. The other three components we found were each focused on a different aspect of the positive symptomatology. The factor named “delusions” was characterized by unusual thought content, stereotyped thinking, preoccupation, delusions, lack of judgment and insight. The “conceptual disorganization” dimension was focused on formal thought disorders, associated with a physical and psychical somatic component, including difficulty in abstract thinking, conceptual disorganization, motor retardation, anxiety, tension, somatic concern. The “paranoid/hostility” factor combined behavioral symptoms with suspiciousness and persecution, namely uncooperativeness, hostility, suspiciousness/persecution, poor impulse control, excitement. We noted that the “hallucinatory behavior” item has not significantly contributed to any of the four components, having a factor load of 0.31, lower than the indicated value as cutoff. “Hallucinatory behavior” also had a low factor load (0.41) in Rapado-Castro study, whereas higher scores, between 0.65 and 0.806, have been found in studies on adult onset FEP [16, 17]. As introduced by DSM 5, the primary symptoms of psychosis are on a continuum, so we can define a gradient of severity between different patients and, over the time, in the same subject [9]. Moreover, populations-based epidemiologic studies have shown that auditory hallucinations are common particularly in young age groups with prevalence rates ranging from 5.7 to 21.0%. For the majority of children and adolescents, these experiences are well below the threshold of clinical disorders and tend to have a benign course [36‐38]. On the other hand, no other clinical phenomena are as likely to be attributed to schizophrenia as hallucinations, yet no other clinical phenomenon common to schizophrenia is as widely spread across numerous other states and disorders, including attention deficit-hyperactivity disorder, conduct disorders, depressive and anxiety disorders, post-traumatic stress disorder, obsessive–compulsive disorder, and borderline personality disorders [36]. It is also notable that hallucinations and delusions, traditionally considered to be evidence of impaired reality testing and gathered together in a single dimension, have been designed better as separate dimensions by DSM-5 [39]. This distinction could be particularly useful in adolescent patients, as suggested by our data in which formal and content thought disorders appeared to be independent of the hallucinatory symptoms. To further support this assumption, Lehembre-Shiah et al. found that only unusual thought content and thought disorganization, not perceptual abnormalities, are associated with conversion to psychosis in clinical high-risk populations, despite the high incidence of perceptual abnormalities [40]. We can argue that, especially at this age range, hallucinations may be considered as transdiagnostic symptoms; thus, the clinical relevance of perceptual abnormalities must take into account markers of severity, comorbid psychopathology and levels of functioning of young patients. This dimensional approach should have a central issue in clinical practice to improve decision making on psychopharmacological and psychosocial interventions in early phase of psychosis. Lastly, compared to the Rapado-Castro model, the solution we obtained does not contain the depressive dimension. This difference may be caused by selection of the sample, since we enrolled patients with non-affective psychosis, while 24.2% of the Rapado-Castro study sample included patients with mood disorders (bipolar and depression) as diagnosis at baseline.

When we used the four factors describe above as a summary of symptoms in cluster analysis, we found that each of these four psychopathological dimensions can identify a discrete clinical subtypes of patients with FEP. This result appears in line with the direction of the DSM-5 that suggests using psychopathological dimension to improve the ability to describe the clinical heterogeneity of schizophrenia, overcoming the traditional DSM-IV subtypes (paranoid, disorganized, catatonic and undifferentiated) [41]. Growing evidences are showing as a dimensional approach to symptoms domains in schizophrenia is more valid and clinically useful in describing clinical picture; in addition a dimensional approach may provide more relevant informations to lead diagnosis and treatment decision [42‐44].

The third aim of this study was to characterize the four clinical subtypes of early onset FEP emerging from the cluster analysis, through the association with sex, age of onset, markers of early neurodevelopment abnormalities and intellectual disability. We found differences between the four groups regarding the early speech and language development and the cognitive profile. Consistent with the neurodevelopmental hypothesis of schizophrenia, premorbid dysfunctions (including language, motor and social deficits) have been well documented in patients with the disorder and were found to be more frequent in subjects that will later on develop schizophrenia during childhood or adolescence, compared to adulthood [4, 45]. The effort of actual and prospective studies is to improve understanding about how psychobiological markers change over time in young people at varying degrees of risk for psychosis and link them to differential clinical, functional or other end points [46]. We found that “delusions” subtype patients had more significant delayed language development than patients in other clinical subtypes. Language dysfunction plays a central role in the clinical presentation of psychosis, but there is a gap in our understanding of the longitudinal development of speech from childhood and of its relationship with the evolution of at risk mental state and psychotic illness [47]. Children with developmental speech/language impairments are at high risk for literacy skills [48]; Hameed et al. found that low performance in literacy skills (spelling, basic real and non-real word reading, reading skills and comprehension), assessed between 7 and 9 age, and a declining pattern over time, was associated with psychotic experiences in early adolescence [49]. Based on our data and of these previous literature findings, we can hypothesize an evolutionary continuity between early alterations of language development, impairment of literacy skills and delusional dimension. Furthermore, a large amount of data suggest that cognitive deficits can be considered as vulnerability markers of psychosis and a core feature of schizophrenia [50, 51]. Some authors suggest that cognitive profiles are quite heterogeneous in patients at FEP indicating quantitative and qualitative differences across the schizophrenia spectrum disorders subgroups [52]. We found that adolescents with cognitive impairment are more likely to develop delusions and conceptual disorganization rather than negative symptoms or paranoid/hostility. This finding could suggest either that the cognitive profile can influence the clinical presentation of psychosis, or that a specific relationship exists between the cognitive impairment and some psychopathological dimensions of psychosis. There is, indeed, considerable evidence for reasoning, attention, metacognition and attribution biases in people with delusions or prone to delusional thinking. The strongest evidence base relates to the “jumping to conclusion” and attributional biases. Recently, these findings have been incorporated into a number of cognitive models that aim to explain delusion formation, maintenance and content within a model of impairment of normal belief formation [53‐55].

Some methodological limitations should be recognized. First of all, regarding the factor analysis we have to consider that the results may be influenced by the measurement instrument and that, according to the items composition of the factors we obtained, slight differences in nomenclature with respect to other studies are been used. In addition, the relatively small sample size would have minimize statistical power and limited our ability to detect subgroup differences. However, the sample was relatively homogeneous for diagnosis, the extracted components showed an adequate internal consistency and the factor loadings of all components were pretty high (>0.5 except for two of them); therefore, the proposed solution can be considered reasonable. Since, at present, contributions on the topic of psychopathological dimension of early onset psychosis are rare, studies on relatively small samples can be worthwhile in contributing to the comprehension of the disorder, although the data are not generalizable. Second, we have to report the lack of inter-rater reliability analyses for the PANSS assessment. Third, although the sample study is antipsychotic-naive, some patients had been exposed to small quantities of benzodiazepines, antidepressants and mood stabilizers, so that the sample cannot be regarded as totally medication-naive.

In conclusion, we presented a four-factor solution describing psychopathological dimensions in childhood and adolescent FEP. “Negative symptoms” was the most robust component; the other three dimensions were “delusions”, conceptual disorganization” and “paranoid/hostility”; hallucinator behavior was not included in any of components. These data could suggest that, among positive symptoms, unusual thought content and thought disorganization may be more distinctive of early onset psychosis than perceptual abnormalities. Moreover, we found that the psychopathological dimensions identified four corresponding clinical subgroups in the study sample, confirming that the dimensional approach may offer a better conceptualization of FEP also in children and adolescent patients. Further replication across additional sample would be needed to confirm this factor solution, in order to generalize the dimensional model to all subjects with early onset schizophrenia spectrum disorders. Finally, this study showed a relationship between the different clinical subtypes and cognitive impairment and speech/language development abnormalities. Longitudinal research designs should be stimulated to capture changing aspects of clinical and neurobiological markers, starting from early stages of neurodevelopment until the onset of overt psychotic symptoms. This evolutive approach would be very useful in early intervention programs to guide risk assessments and treatment decisions of children and adolescents at risk of psychosis.