Depression and Schizophrenia: Sleep, Medical Risk Factors, Biomarkers, and Treatment

Insulin resistance and metabolic syndrome are well known to be associated with psychiatric disorders, most commonly due to the long-term administration of antipsychotic medication as well as to maladaptive lifestyle choices. Watson and colleagues (3) use data from the Netherlands Study of Depression and Anxiety cohort to explore the extent to which insulin resistance and prediabetes may be risk factors for the later development of major depression. In this study, the investigators analyzed longitudinal data from 601 adults that did not have a history of either depression or anxiety disorders to explore the hypothesis that surrogate markers of insulin resistance (high triglyceride-HDL ratio, fasting glucose, and waist circumference) would predict the development of major depression. Over the 9-year follow-up period, 14% of the cohort developed major depression, and all the surrogate markers were associated with an increased risk of developing major depression. The largest effect was observed for high triglyceride-HDL ratios, reflected by a 1.89 hazard ratio. The authors also examined the consequences of developing prediabetes (high fasting plasma glucose level) over the first 2 years of the study and found a hazard ratio of 2.2 for the likelihood of developing major depression over the next 7 years. These findings are important as they move beyond the well-known cross-sectional association between insulin resistance and psychopathology, providing further evidence that insulin resistance is a risk factor for the development of depression. In interpreting the findings, it is important to keep in mind that surrogate markers of insulin resistance were used in lieu of the direct assessment of insulin resistance, which is accomplished with the more complicated euglycemic clamp test. While indirect and perhaps not as specific, the surrogate measures highlighted in this paper are readily obtainable in clinical practice. Dr. Roger McIntyre from the University of Toronto contributes an editorial that discusses the findings and speculates about potential causal mechanisms that might underlie the linkage between preexisting insulin resistance and the development of depression (8).

Lyme disease, caused by a Borrelia bacteria and transmitted by deer ticks, is very common, with an estimated 476,000 individuals annually developing this illness in the United States (9). Numerous reports have described a relationship between Lyme disease and psychiatric symptoms, and Fallon et al. (4) add convincing temporal evidence to support this contention. In this study, the investigators used Danish registry data, which included more than 6 million individuals, to identify 12,156 individuals, 3 years of age or older, who were diagnosed with Lyme disease. Importantly, these individuals did not have a previous history of a mental disorder or suicide attempt. Overall, the results demonstrated that Lyme disease was associated with an increased rate of any mental disorder, affective disorder, or suicide attempt. While the rate of mental disorders in the Lyme disease group was increased by 28%, it is important to keep the actual numbers in mind, as relatively few individuals with Lyme disease actually developed a mental disorder (733 per 100,000 person-years associated with Lyme disease versus 567 per 100,000 person-years in the non-Lyme disease group). Notably, the greatest relative risk for developing any mental disorder occurred within the first 6 months after diagnosis (96% increase). Furthermore, the relative rate of developing any mental disorder increased in relation to additional episodes of Lyme disease. For affective disorders there was a 42% increase in the Lyme disease group compared with individuals without Lyme disease; for suicide attempts a significant 2.45 increased rate was noted beginning 3 years after diagnosis. Of note, the investigators also examined the extent to which developing Lyme disease-associated neurological symptoms, so-called neuroborreliosis, further increased the risk of developing mental disorders and did not find any such association. The findings from this study document the increased risk of developing psychiatric symptoms and engaging in suicide attempts in individuals with Lyme disease and are consistent with other work linking various infections and inflammation to increased rates of psychiatric illness.

Weiser et al. (5) use VA pharmacy data from 2010 to 2015 to compare the use of different oral and long-acting antipsychotics in schizophrenia patients in relation to discontinuation rates and numbers of hospitalizations. The authors argue that discontinuation rates of antipsychotics can be used as a proxy for their efficacy and tolerability. The data used in this study were from 37,368 outpatient veterans with schizophrenia, and the antipsychotic medication treatment comparisons were made relative to individuals treated with olanzapine. Regarding discontinuation rates, oral clozapine, long-acting injectable formulations of aripiprazole, paliperidone, and risperidone—as well as the concomitant use of two antipsychotics—proved to be superior to oral olanzapine treatment. The largest effect observed was for clozapine, which compared with olanzapine had a 0.43 hazard ratio. In contrast, treatments that were significantly more likely to be discontinued sooner than oral olanzapine included first-generation antipsychotics as well as oral risperidone, aripiprazole, ziprasidone, and quetiapine. These findings are generally consistent with other recent studies, including a large meta-analysis recently published in the Journal (10) that demonstrated better “acceptability” of long-acting antipsychotics. Interestingly, when assessing the risk of hospitalizations in the current study, none of the medications performed significantly better than oral olanzapine. However, several agents were associated with a significantly greater risk of hospitalization (oral first-generation antipsychotics, long-acting paliperidone, long-acting haloperidol, long-acting fluphenazine, oral lurasidone, oral quetiapine, and oral ziprasidone), although it remains unclear why this was the case. In their editorial (11), Dr. Stefan Leucht from the Technical University of Munich and Dr. Jari Tiihonen from the Karolinska Institute provide a comprehensive overview of previous findings commenting on the consistency of the current findings in relation to previous work. In addition, they make the point that while better for improving adherence, long-acting antipsychotic formulations are not pharmacologically or inherently more effective than their shorter acting counterparts. They also speculate as to how the study design in the paper by Weiser et al. may relate to the authors’ failure to find an expected reduction in hospitalizations with clozapine and long-acting antipsychotic formulations.

Posselt et al. (6) used data from Danish registries to assess the standard clinical use, or real-world efficacy, of the OPUS intervention for patients experiencing their first psychotic episode. OPUS is a 2-year intervention that integrates assertive treatment with family involvement, social skills training, and work with other mental health providers, including psychiatrists. The data from the standard clinical use of OPUS was compared with results from a previous randomized controlled trial in which OPUS was compared with treatment as usual. As a rationale for the current study, the authors point out the importance of demonstrating efficacy in real-world settings to establish the generalizability of findings from research to clinical practice. The original OPUS trial, conducted from 1998 to 2001, demonstrated significant effects on reducing psychotic and negative symptoms and substance abuse, as well as on improving the functional capacities of patients. Somewhat surprisingly, the results from the current comparison demonstrated greater effects of OPUS when used in the real-world setting compared with its use in the clinical trial. For example, mortality rates were lower in the real-world treatment group compared with that in the clinical trial, as were the number of psychiatric admissions and bed days. Additionally, 5 years after the 2-year OPUS treatment, the real-world patients were more likely to be working compared with patients in the clinical trial. The authors emphasize an important limitation in interpreting this study, which is that the real-world OPUS treatment occurred well after the clinical trial. Thus, it is conceivable that the observed increased efficacy in the real-world use of OPUS could in part be due to overall changes in psychiatric and medical treatment associated with the accepted clinical use of OPUS in Denmark. Nonetheless, the findings demonstrate that the efficacy of the OPUS intervention was not diminished when used in a clinical setting and support the widespread use of such a systematic intervention to enhance the outcomes of individuals experiencing their first psychotic episode. In his editorial (12), Dr. Tyrone Cannon from Yale University discusses the findings in relation to other studies of early intervention services. While acknowledging their important value and efficacy during the active treatment phase, he points out that OPUS and similar interventions have diminished efficacy over the long term.

The auditory oddball paradigm has been extensively used to assess neurophysiological brain responses to the presentation of deviant stimuli, and alterations in this response have been demonstrated across various psychiatric illnesses. In this paradigm, quantitative EEG is used to measure the neural response to expected and unexpected auditory stimuli and specific event-related potentials, such as the P300. The signal for the P300 is strongest over the parietal lobe and is thought to reflect broad cortical and subcortical synchronization that is linked to processes involving orienting, attention, and responses to novelty. The most consistent findings have been in patients with schizophrenia documenting decreased amplitude of the P300 response. In this issue of the Journal, Parker et al. (7) used the oddball paradigm to identify EEG features that distinguished diagnostic groups from each other. The sample studied was from the Bipolar-Schizophrenia Network for Intermediate Phenotypes and included individuals with diagnoses across the schizophrenia-bipolar spectrum: schizophrenia (N=597), schizoaffective disorder (N=225), bipolar disorder with psychosis (N=171), bipolar disorder without psychosis (N=66), and healthy subjects (N=415). The results demonstrated that numerous parameters of the oddball-related EEG responses differed among the diagnostic groups. Specifically, features of the P300, as well other event-related potential measures such as the N200, varied across the diagnostic continuum: healthy > bipolar disorder without psychosis/bipolar disorder with psychosis > schizoaffective disorder/schizophrenia. Interestingly, across the entire study cohort these alterations were also related to the transdiagnostic assessment of cognitive abilities. It is important to note that these findings generally replicate previous findings from an earlier study from the Bipolar-Schizophrenia Network for Intermediate Phenotypes (13). In addition to these findings, the authors showed that the P50 response distinguished patients with bipolar disorder without psychosis from those with psychosis, and the P200 and N200 responses distinguished patients in both bipolar groups from those with other diagnoses. In his editorial (14), Dr. Daniel Javitt from Columbia University provides a historical perspective on the use of EEG measures in psychiatry, emphasizes the validity and replicability of the findings from the current study, and discusses the use of measures like P300 as potential biomarkers for better defining patient subgroups and in developing new treatments.

Science and research are iterative processes that, over time, have the potential to result in significant changes in patient care and clinical practice. The papers in this issue of the Journal report findings that represent significant steps in this iterative process by extending and validating salient results from previous studies relevant to clinical psychiatry. The overviews on sleep in depression and sleep in schizophrenia portray the advancements in the use of sleep EEG measures in psychiatry. These overviews also lay the foundation for future studies aimed at using an understanding of illness-specific altered sleep physiology to inform the development of new treatment strategies. Other new findings of importance presented in this issue include longitudinal data supporting an increased risk of developing major depression that is conferred by preexisting measures of insulin resistance as well as an increased risk of developing affective disorders and engaging in suicide attempts after developing Lyme disease. Related to schizophrenia, data from the large VA registry points to decreased discontinuation rates with some long-acting antipsychotic formulations and clozapine. These data should further encourage clinicians to increase their usage of long-acting antipsychotic formulations and clozapine in treating schizophrenia. Data from the large Danish registry emphasizes the real-world value of systematic, integrated psychosocial and psychiatric interventions for treating first-episode psychoses. Finally, data from the Bipolar-Schizophrenia Network for Intermediate Phenotypes demonstrates the validity and replicability of using auditory evoked response potentials as potential biomarkers for subtyping patients along the bipolar-schizophrenia continuum and as an intermediate brain phenotypic marker to guide the evaluation of new treatments.