Bipolar Disorder Vulnerability: Perspectives from Pediatric and High-Risk Populations

By Jair Soares and Paolo Brambilla

Bipolar Disorder Vulnerability: Perspectives from Pediatric and High-Risk Populations synthesizes our current understanding of high-risk and pediatric populations to aid readers in identifying markers of vulnerability for the development of bipolar disorder, with an ultimate goal of the development of drug targets and other therapies for early diagnosis and treatment. The book provides readers with an understanding of biological and environmental factors influencing disease manifestation that will aid them in defining discrete clinical stages and, importantly, establish an empirical basis for the application of novel therapeutics in a phase of illness during which specific treatments could more effectively alter disease course.

Whereas most of the literature available on the pathophysiological mechanisms of bipolar disorder focuses on chronically ill adult individuals, this represents the only book that specifically examines pediatric and high-risk populations. An estimated 30 to 60 percent of adult bipolar disorder patients have their disease onset during childhood, with early-onset cases representing a particularly severe and genetically loaded form of the illness.

• Highlights diverse translational methodologies, including functional and structural neuroimaging, neuropsychological testing and integrated genomics
• Examines molecular trajectories in youth with bipolar disorder and unaffected youth at high risk for developing bipolar disorder
• Explores the interaction between genomic and environmental influences that shape behavior

• Language: English
• Publisher: Academic Press
• Release date: Jun 9, 2018
• ISBN: 9780128125601

Book preview

Spain

Chapter 1

The bipolar prodrome

Danella M. Hafeman; Boris Birmaher    Department of Psychiatry, School of Medicine, University of Pittsburgh, Pittsburgh, PA, United States

Abstract

Converging evidence suggests that the onset of bipolar disorder is usually preceded by a prodromal period lasting months to years. In this chapter, we review studies that point to this prodrome, including family risk studies, studies of individuals with depression or cyclothymia, community studies, and retrospective studies. We review assessments used to predict bipolar disorder, as well as a risk calculator for predicting new-onset bipolar disorder. While each study design has its limitations, taken together, the literature begins to point to an initial nonspecific prodrome, including symptoms such as anxiety, depression, and mood lability. Closer to disorder onset, subthreshold manic symptoms become more prominent, and characterize a more specific prodrome. A better understanding of a staged prodrome is crucial to inform clinical management of individuals at familial or symptomatic risk of bipolar disorder, and aid in the selection of an ultra-high-risk population for future studies of biomarkers and therapeutics.

Keywords

Bipolar disorder; Prodrome; At-risk; Risk calculator; Study design; Questionnaires

Chapter Outline

Introduction

Methods

Study Design: Challenges and Strategies

Questionnaires/Assessments

Results

Family Studies

Depressed Samples

Cyclothymia/BD-NOS Samples

Community Samples

Retrospective Studies: Metaanalyses and Reviews

Staging Models

Risk Calculator

Summary and Future Directions

References

Introduction

It has long been recognized that bipolar disorder rarely comes out of the blue, but rather is usually preceded by a period of time, lasting at least a month and up to several years of prodromal symptoms (Correll, Hauser, et al., 2014). A better understanding of prodromal bipolar disorder is essential for several reasons. First, there are often delays in diagnosis and treatment of bipolar disorder lasting an average of 10 years, leading to increased morbidity and poor function (Lish, Dime-Meenan, Whybrow, Price, & Hirschfeld, 1994). Second, even when individuals are identified early, it is often unclear how to treat them, due to concern that some medications (such as antidepressants) might exacerbate their symptoms. A better understanding of the bipolar prodrome might allow us to identify ultra-high-risk individuals, conduct studies to understand the neural correlates of bipolar risk, and test the effects of various classes of medications and/or psychotherapies in this population. In this chapter, we review what is currently known about the bipolar prodrome, from both retrospective and longitudinal studies. We first describe the methods that have been used to assess the bipolar prodrome, including study design and assessment tools. Next, we describe findings from individual studies that shed light on the course and characteristics of this prodrome. Finally, we describe recent directions to integrate these findings (e.g., staged model, risk calculator), and we present a current model for the prodrome based on the extant literature.

Methods

Study Design: Challenges and Strategies

There are significant challenges with studying prodromal symptoms of a relatively rare disorder like bipolar disorder. It would simply be impractical to well-characterize a sample from the general population, and follow them long enough to have a sufficient number develop bipolar disorder. Thus, there are a variety of strategies that have been used to shed light on this topic, each with its own strengths and limitations. Each provides a different view on this prodromal period, although, encouragingly, the findings from different strategies are quite similar, as we will discuss later. What we know about the bipolar prodrome comes from the following types of studies:

•Family studies: There are several longitudinal studies of offspring of parents with bipolar disorder, which have well-characterized these at-risk offspring in childhood and adolescence, and prospectively followed them to assess new-onset disorders, including bipolar disorder (Birmaher, Axelson, Monk, et al., 2009; Duffy, Alda, Crawford, Milin, & Grof, 2007; Egeland et al., 2003; Hillegers et al., 2005). Because the onset of bipolar disorder is much more prevalent in individuals at familial risk, there are sufficient converters to evaluate clinical characteristics that might precede new-onset disorder. In addition, these individuals are at higher risk of other disorders as well, and have high levels of subsyndromal symptoms, so they also represent a clinically at-risk population that is important to characterize and better understand. There are some limitations to this approach, however. First, we don’t know that the course of the bipolar prodrome is similar in individuals with vs. without a first-degree relative with bipolar disorder, so it is unclear the degree to which these findings are generalizable to individuals without such family risk. Second, the prevalence of syndromal bipolar disorder (bipolar-I/II) in offspring is still fairly low (e.g., 8.4% in BIOS, though not all participants have passed the risk period), so large samples are required to have sufficient new-onset cases to make inference about predictors (Axelson et al., 2015). To handle this limitation, some groups have instead assessed less stringent outcomes, including bipolar spectrum disorder (which includes bipolar disorder, not otherwise specified) and mood disorder (which includes unipolar or bipolar depression). Another approach is to identify high-risk samples within the offspring of parents with bipolar disorder, based on the presence of mood, anxiety, and/or mood lability symptoms. Third, there is the important issue of comorbidity in the bipolar parents, and whether differences observed in offspring are related to the family history of bipolar disorder, per se, or a comorbidity. For example, in the Pittsburgh Bipolar Offspring Study, Attention-Deficit/Hyperactivity Disorder (ADHD) was higher in at-risk offspring than community controls. However, after adjusting for confounders (including nonbipolar psychopathology in both biological parents), this difference was no longer significant (Birmaher, Axelson, Monk, et al., 2009). Studies that recruit healthy controls (as opposed to including parents with nonbipolar psychiatric disorders) cannot necessarily conclude that a particular difference in at-risk vs offspring of healthy parents is due to the bipolar disorder (vs higher rates of ADHD, for example). Fourth, another critical issue when carrying out family risk studies is blinding. If the interviewer knows that a parent has bipolar disorder, ratings might be elevated due to expectations of worse outcomes. Most studies discussed here were blinded, except for the Dutch study, which only included offspring of parents with bipolar disorder (Mesman, Nolen, Reichart, Wals, & Hillegers, 2013). Fifth, when using parent report to assess a child’s psychiatric symptoms, it is important to take into account the current mood state of the reporting parent, since this can impact symptom ratings (Maoz, Goldstein, Goldstein, et al., 2014). This is especially crucial in family risk studies where, by definition, at least one parent has bipolar disorder, and thus over-reporting of symptomatology could bias parent-report measures of child psychopathology. Sixth, depending on the age range, participants might not have reached the peak period of conversion to bipolar disorder; thus, there is the possibility that some of the nonconverters might still develop the disorder. These issues have been summarized in previous reviews (DelBello & Geller, 2001; Hauser & Correll, 2013; Hunt, Schwarz, Nye, & Frazier, 2016).

•Unipolar depression studies: An episode of unipolar depression, particularly with earlier age of onset and psychotic features, sharply increases the risk of new-onset bipolar disorder (Akiskal, Maser, Zeller, et al., 1995; Kovacs, 1996; Strober & Carlson, 1982). Thus investigators have prospectively assessed depressed individuals, to determine symptom predictors of conversion from unipolar depression to bipolar disorder. The strengths of this approach are the fact that it is prospective, and using a select population with conversion rates that allow for adequate cases of new-onset bipolar disorder, at least over a long follow-up (e.g., 19.6% over a mean follow-up period of 17.5 years) (Fiedorowicz et al., 2011). However, there are also some limitations. First, the rate of conversion is still relatively low over shorter periods of follow-up, thus necessitating a longer follow-up period or larger sample for adequate converters. One way that investigators have handled this is to narrow the selection criteria to participants with depression with psychotic features, thus increasing the base rate for developing bipolar disorder; however, this also makes recruitment more difficult. Second, as with the family studies, there is also the possibility that participants who have not passed the peak age of conversion might be misclassified as nonconverting. Third, selection of a sample based on unipolar depression means that the results might be specific to individuals that debut with a major depressive episode, and not necessarily generalize to those who have a different presentation (e.g., cyclothymia or initial mania/hypomania).

•Cyclothymia/bipolar disorder-not otherwise specified (BD-NOS) samples: One of the strongest predictors of new-onset bipolar disorder is subthreshold manic episodes (see below for specific findings), and thus several studies have assessed the clinical variables that predict progression from BD-NOS to BD-I/II (Akiskal, Djenderedjian, Rosenthal, & Khani, 1977; Alloy, Urošević, et al., 2012; Axelson et al., 2011). Significant strengths of this approach are that conversion is high (30%–50% in 18 months to 5 years; see below for details), meaning that even small studies have enough power to assess prospectively the impact of other clinical variables on progression. One possible limitation, discussed in detail later, is that these rates of conversion call into question whether BD-NOS (or cyclothymia) is a precursor to disorder, or part of a bipolar spectrum disorder; thus it is unclear whether assessing characteristics of individuals with BD-NOS really constitutes studying predictors (vs correlates) of disorder, and/or predictors of disorder progression.

•Community samples: Several epidemiologic studies have assessed the effect of subsyndromal symptoms, including symptoms of depression, mania, and psychosis, on the onset of bipolar disorder (Homish, Marshall, Dubovsky, & Leonard, 2013; Regeer et al., 2006; Tijssen et al., 2010b). These studies have large numbers (generally > 2000 subjects) over a long follow-up to yield an adequate number of new cases of bipolar disorder. The trade-off in these types of studies is that there is generally infrequent follow-up and diagnoses are usually based on abbreviated assessments (often without review by a psychiatrist). Thus, while there is strength in numbers, and in the generalizability of findings to a community sample, a fine-grained assessment of each individual participant is usually lacking. Instead of a full diagnostic interview, the Composite International Diagnostic Interview (CIDI) is often used; however, this instrument has shown excellent concordance with more comprehensive measures (Kessler et al., 2013).

•Retrospective: These studies interview individuals with bipolar disorder, often during the first episode of disorder (though not always), about the nature and timing of symptoms that they were having prior to the onset (Correll, Hauser, et al., 2014; Egeland, Hostetter, Pauls, & Sussex, 2000). One strength of retrospective studies is that they can be carried out with minimal resources, and do not require follow-up of the sample; thus it is feasible to collect multiple samples to confirm findings. Also, unlike prospective studies (see below), the sample is not restricted based on criteria such as family history or an episode of depression. In this way, retrospective studies assess all comers and are perhaps more generalizable to the population that will develop bipolar disorder. However, there is a major limitation, which is that these findings are based on the recall of subjects who have already developed the disorder. Individuals with bipolar disorder might have a more distinct memory of subsyndromal mood symptoms during adolescence, for example, than someone who did not go onto develop bipolar disorder; this could lead to recall bias, and inflate the association between previous symptoms and the development of bipolar disorder. Even in the absence of such bias, subjects in retrospective studies are often asked to recall details about experience more than a decade before; this may lead to incorrect or incomplete reporting of previous symptoms, especially in someone with neurocognitive deficits (as often seen in bipolar disorder). An alternative method is to use records prior to bipolar disorder onset; this removes the problem of recall bias, but clinical records are often incomplete (Egeland et al., 2000). Also, depending on where the bipolar sample is recruited from (e.g., inpatient unit), the reported prodromal symptoms might not be generalizable to individuals with bipolar disorder who do not present to this setting. Thus retrospective studies provide an important source of clinically rich information about what the bipolar prodrome might look like, but these results should be confirmed using additional methods.

In summary, each of these study designs has important strengths, but also significant flaws that might bias findings and/or limit generalizability of results. Thus the strongest conclusions will come from observing similar findings across different study designs.

Questionnaires/Assessments

Based on a growing knowledge base about the symptoms that predict new-onset bipolar disorder, several scales and instruments have been developed to characterize better this risk. Many of these were reviewed recently (Ratheesh, Berk, Davey, McGorry, & Cotton, 2015); while a few of these questionnaires showed promise in a single study, e.g., the General Behavioral Inventory (GBI) and the Manic Symptom Subscale of the Child Behavioral Checklist (CBCL-MS), none was replicated in high-quality studies. Scales used to screen for bipolar disorder in youth, a closely related though not identical problem, have been evaluated elsewhere (Youngstrom et al., 2004). These authors found that, in general, parent report of manic symptoms better distinguished youth with bipolar disorder from healthy controls than either youth or teacher reports. Here, we discuss scales and assessments that have been used in the attempt to predict onset of bipolar disorder. The list below is meant to be not a comprehensive review of questionnaires that could potentially predict bipolar disorder, but rather an overview of strategies that investigators have used to estimate the risk of conversion, and the degree to which these have been validated in prospective studies.

•Child Behavior Checklist (CBCL) subscales (parent-report): One longitudinal study found that a high score on CBCL subscales of attention, aggression, and anxiety/depression predicted new-onset bipolar disorder in youth with ADHD (Biederman et al., 2009); thus they termed this the pediatric bipolar disorder subscale. However, these authors and others found that high scores on this scale predicted not only bipolar disorder, but also other disorders such as depression and conduct disorder (Diler et al., 2009; Meyer et al., 2009); the scale also predicted severity of disorder and poor function, and thus seems to be an indicator of general psychopathology. As such, it is now more often called the dysregulation profile (Althoff, Verhulst, Rettew, Hudziak, & van der Ende, 2010). More recently, Papachristou et al. (2013) developed the CBCL mania scale (CBCL-MS), based on 19 items from the CBCL (Table 1); the scale was found to have high internal consistency, and to discriminate between youth with BD-I and healthy controls (AUC = 0.64) (Papachristou et al., 2013). Youth with BD-I also had higher scores on the CBCL-MS than youth with anxiety (P = .004) and major depressive disorder (P = .002), but not compared to youth with ODD or ADHD. In a longitudinal community study of Dutch adolescents, the authors found that those in mildly and highly symptomatic classes (based on their CBCL-MS scores at age 11) were at a twofold and fivefold risk, respectively, to develop new-onset bipolar disorder by the age of 19 (Papachristou et al., 2017). After adjustment for confounders, this scale was not predictive of new-onset anxiety or depression, though those in the highly symptomatic class were more likely to have diagnoses of ADHD, oppositional defiant disorder (ODD), and conduct disorder.

Table 1

BD, Bipolar Disorder; CARE, Child and Adolescent Research Evaluation; CAS, Child Assessment Scale; CBCL, Child Behavioral Checklist; DIGS, Diagnostic Interview for Genetic Studies; EAS, Emotionality Activity Sociability Survey; ECI-4, Early Childhood Inventory-4; K-SADS-PL, Kiddie-Schedule for Affective Disorders and Schizophrenia for School-Aged Children; LiNR, Offspring of lithium-nonresponsive parent; LiR, Offspring of lithium-responsive parent; MD, Major Depressive Disorder.

•Bipolar Prodrome Symptom Interview and Scale (BPSS clinician-administered): Correll et al. (2007) initially developed a retrospective version of this scale, which included 36 items that assessed subthreshold symptoms of mania, depression, and psychosis. In 52 individuals with child- or adolescent-onset mania, the authors found that all participants had experienced at least one moderately severe manic symptom prior to onset. While approximately half had an insidious onset (> 1 year of symptoms), most of the remainder had a subacute onset (1 month to 1 year); only a small minority (3.8%) had less than a month of symptoms prior to onset. Most common symptoms were subthreshold manic symptoms (irritability, racing thoughts, and increased energy) and depressed mood. The prospective version of this scale (BPSS-P) was developed more recently, and has also been shown to discriminate well between bipolar disorder, other psychopathology, and healthy controls, with expected correlations with other scales of mania and depression (Correll, Olvet, et al., 2014). The prospective utility of this scale has not yet been