The significance of at-risk or “prodromal” symptoms for bipolar-I disorder in children and adolescents

Abstract

Background

While in the early identification and intervention of psychosis, specific instruments and risk criteria have been generated and validated, research into preventive strategies for bipolar-I disorder (BP-I) has only recently gained momentum. Since first signs of BP-I often start before adulthood, such efforts are especially important in the vulnerable pediatric population.

Methods

Narrative review of data regarding the presence and nature of potentially prodromal, i.e., sub-syndromal, symptoms prior to BP-I, defined by first-episode mania, focusing on pediatric patients.

Results

The possibility of early identification of youth at high-risk for BP is supported by retrospective studies of BP-I patients, and prospective studies of community samples, offspring of BP-I subjects, youth with depressive disorders, and patients at high-risk for psychosis or with BP-spectrum disorders without lifetime history of mania. These data provide essential insight into potential signs and symptoms that may enable pre-syndromal identification of BP-I in youth. However, except for offspring studies, broader prospective approaches that focus on youth at risk for BP-I and on developing specific interviews/rating scales and risk criteria are mostly missing, or in their early stage.

Conclusions

More work is needed to determine valid clinical high-risk criteria, to distinguish risk factors, endophenotypes and comorbidities from specific prodromal symptomatology, and to develop phase-specific interventions that titrate the risk of intervention to the risk of transition to mania and to functional impairment or distress. Moreover, studies are needed that determine potential differences in prodromal symptoms and trajectories between children, adolescents and adults, and the best phase-specific interventions.

Introduction

Bipolar I disorder (BP-I) markedly impairs behavior, social, occupational and physical functioning, and general health^1–3. For the pediatric subtypes, adverse consequences are even more severe, including higher rates of treatment refractoriness, psychiatric comorbidities and functional disability^4–8. A major challenge for improved outcomes is the commonly large lag between symptom onset and first correct BP-I diagnosis and appropriate treatment^4,9. Although the typical illness onset according to the Diagnostic and Statistical Manual (DSM)-IV is at around age 20 years, 50–67% of individuals with BP-I report first illness signs prior to age 18^5,9. Moreover, prepubertal onset is also much more common than in non-affective psychotic disorders¹⁰. All of the features mentioned above make BP-I a prime condition for early identification and intervention.

However, despite the urgency to develop rating instruments and criteria for the prodromal, i.e., sub-syndromal and sub-threshold presentations of BP-I, which is required to validate risk criteria in prospective studies and test early interventions, as has been done successfully in psychosisresearch^11–14, these developments have been lagging behind in BP-I research until fairly recently^15–23.

In this paper, we will review several lines of evidence supporting the existence of a BP-I prodrome, i.e., a sufficiently long phase of beginning symptoms foreshadowing a first mania episode that would allow for early identification and intervention in a sufficiently large cohort of patients. We focus on data and the application in youth, discussing also conceptual and practical challenges.

Methods

In this narrative review, we i) review conceptual and practical problems in defining a BP-I prodrome, particularly in youth, ii) summarize the evidence for the presence of a subthreshold, symptomatic prodrome preceding BP-I, and iii) evaluate the current knowledge highlighting the consequences for research and clinical care of youth at possible risk for BP-I.

Although, retrospectively, depression may have been the first illness “episode”, the onset of BP-I is defined for the purpose of this review diagnostically (rather than biologically, for which data are currently absent) as the onset of a first manic or mixed episode. Moreover, although it is clinically relevant to identify precursor and risk states for BP-II or BP not otherwise specified (NOS), we focus almost exclusively on BP-I, as, to date, data are not available to adequately characterize the prodrome even preceding full mania.

For the purpose of this review and concordant with the psychosis field¹², we define prodromal or subsyndromal symptoms and signs as symptomatic illness manifestations that are attenuated versions of a full mania syndrome. In this context, hypomania, cyclothymia and BP-II can all be seen as the latest, most proximal “prodrome to BP-I. We further define risk factors as biological, environmental, behavioral, emotional, cognitive or psychological states and events that increase the likelihood of developing BP-I, but that are not thought to be part of the symptomatic evolution towards fulfilling current diagnostic criteria. We finally define comorbidities as syndromal manifestations of non-affective disorders that are distinctly classified under current nosological conventions, even if there may be symptomatic overlap with threshold or subthreshold mania symptoms. We realize that these definitions are preliminary and will likely be revised given new knowledge.

Results

In the past few years, a number of reviews assessed the scarce retrospective and even scarcer prospective literature on prodromal symptoms and signs in adults with BP-I^{15–18,21–25}. Even more recently, several articles focused specifically on children and adolescents^19,20,26. Across these reviews, the lack of specificity of prodromal symptoms has been bemoaned. The authors cautiously concluded that there is some evidence that prodromal states precede BP-I, including in children and adolescents, at least in a sizeable subgroup of individuals, but that more longitudinal research including control groups is needed to establish the definition of the BP-I prodrome and to ultimately enable early identification of BP-I.

Conceptual and Practical Difficulties with the BP-I Prodrome

While the existence of a psychotic prodrome is well documented and validated with criteria that have yielded transition rates to psychosis within 6 months to 3 years of 10–40% in criteria-based at-risk samples, which are high enough to test preventive interventions^11,12, the existence of a sufficiently specific BP-I prodrome is still debated.

A major challenge for identifying a BP-I prodrome, particularly in youth, is the complex phenomenological definition and course of BP. The DSM-IV-TR²⁷ does not distinguish between adults and youth. However, BP-I youth seems to differ from adults in several ways, including the onset pattern, symptom and cycling characteristics, as well as inter-episode functioning (Table 1)^28–31. Interestingly, Masi and colleagues¹⁰ even report differences between childhood and adolescent onset patients. While childhood-onset (before age 12 years) BP-I was more frequently associated with male gender and co-morbid attention deficit-hyperactivity disorder (ADHD) and/or oppositional defiant disorder (ODD), and a rather subcontinuous course (41.2% episodic), adolescent-onset BP-I was more frequently associated with an episodic course (76.8%), indicating also potential differences regarding prodromal states. The frequent comorbidity of BP with ADHD and ODD in youth is complicated by the fact that certain symptoms overlap and can be rated for both conditions (see Table 2)³¹. The large “symptomatic/criteria” overlap makes it the more important to consider only B criteria mania symptoms when they are related to A criteria mood symptoms. However, this can be difficult and the overlap of mania criteria with other conditions and the less episodic course in youth has triggered ongoing debate about whether pediatric and adult BP-I are the same conditions, or whether pediatric BP-I may be a developmental subtype or misdiagnosed due to overlapping symptoms with ADHD, ODD, substance use/experimentation, irritable depression or life crises^15,31,32. The current definition of mania constitutes the key aspect in this debate. Pediatric mania has been characterized by very rapid, brief, recurrent episodes, either lasting hours to a few days, or constituting the overall impaired baseline functioning^29,33,34. Accordingly, Axelson and colleagues³⁵ reported that the short duration of episodes in youth, which are shorter than DSM IV criteria require, was the main reason why they were diagnosed with BP NOS instead of BP-I. Furthermore, depending on whether elated mood or irritability is the leading A criterion symptom, as many as an additional 3 or 4 BP-I criteria symptoms at suprathreshold severity are required for a diagnosis of mania²⁷. This leaves the possibility of diagnostically still subsyndromal/prodromal stages, despite the presence of either 3 or 4 full severity symptoms that last for ≥1 week.

Table 1.

Comparison of Characteristics in Bipolar I Disorder with Onset in Childhood and Adolescence Compared to Onset in Adulthood ^10,29–31

Characteristic	Child and Adolescent Onset	Adult Onset
Family History of BP-I	Common, especially in prepubertal onset form	Not as common
Onset Pattern	Often more protracted	Often more abrupt
Cycle Length	Often short, frequent ultra-rapid and ultradian cycling (especially in prepubertal-onset form)	Often more demarcated episodes lasting week(s), rather than hours or days
Inter-episode Interval	Often impaired	Asymptomatic intervals not uncommon
Euphoria	Irritability > euphoria	Euphoria not uncommon
Mixed symptoms	Very common	Not as common
Psychosis	Common	Not as common
Substance Abuse Comorbidity	Less common	More common
ADHD Comorbidity	Common, especially in prepubertal-onset form	Uncommon
Aggression and Impulsivity	Common, especially in prepubertal-onset form	Less common

ADHD: attention deficit-hyperactivity disorder; BP-I: bipolar I disorder

Table 2.

Symptomatic Overlap of Criteria for Attention Deficit-Hyperactivity Disorder and Oppositional Defiant Disorder with Diagnostic Criteria for Mania

DSM-IV ADHD Criteria Mapping on to Mania Criteria		DSM-IV ODD Criteria Mapping on to Mania Criteria
ADHD Criteria	Overlapping Mania Criteria	ODD Criteria	Overlapping Mania Criteria
Inattention Criteria		Is often spiteful and vindictive	-
Does not follow through	-	Often loses temper	Irritable mood
Not organized	-	Often argues with adults	Irritable mood, grandiosity
Loosing things	Distractibility	Often actively defies or refuses to comply with adults’ requests or rules	Grandiosity, excessive involvement in pleasurable activities that have a potential for painful consequences
Carelessness	Excessive involvement in pleasurable activities that have a potential for painful consequences	Often deliberately annoys people	Irritable mood, grandiosity
Inattention	Distractibility	Often blames other for his/her mistakes or misbehavior	Irritable mood, grandiosity
Not Listening	Distractibility	Often is touchy or easily annoyed by others	Irritable mood
Avoiding school work	Excessive involvement in pleasurable activities that have a potential for painful consequences	Is often angry and resentful	Irritable mood
Distractible	Distractibility
Forgetful	Distractibility
Hyperactivity/Impulsivity Criteria
Cannot play quietly	-
Cannot wait turn	[Impulsivity is a common associated symptom]
Fidgets	Psychomotor agitation
Leaves seat	Psychomotor agitation
Runs / climbs excessively	Psychomotor agitation
“On the go”	Increased goal-directed activity or psychomotor agitation
Talks Excessively	Pressured speech, flight of ideas
Blurts out answers	Pressured speech
Interrupts	Pressured speech

Bolded symptoms can be scored by the presence of symptoms of the other disorder

Further complicating the delineation of mood states, both prodromal and subsyndromal, is the fact that, in youth, irritability is both an A criterion for depression and for mania ²⁷, and that mixed episodes are frequent³⁶. Finally, BP-I may begin with depressive episodes or with a manic/mixed episode. This fact adds further complexity compared to the psychosis prodrome, which generally is much less episodic and for which presence of only one single attenuated positive symptom is sufficient. Moreover, since mood and behavioral activation are not uncommon and quite non-specific in childhood and adolescence, they may co-occur in patients at risk for psychosis, but they are not defining symptoms of the attenuated or full, syndromal illness expression, as in BP-I.

Taken together, these general and developmental characteristics complicate the delineation of normal and abnormal mood states and the attribution of symptoms to a true BP-I prodrome versus comorbid or unrelated conditions, crises or extremes of normal development. However, it remains to be seen if and to what degree these mood episodes or comorbid conditions are risk or vulnerability markers for the future development of BP-I, even if they may not (yet) represent subsyndromal illness expressions. A delineation of risk markers versus prodromal symptoms will require much more research as will the discovery of reliable endophenotypes and biomarkers of illness and illness development.

Evidence for a Presence of a Bipolar–I Disorder Prodrome in Youth

Below, we summarize studies in various populations investigating the identification of a clinical high-risk state for BP-I, focusing especially, but not exclusively, on pediatric data.

Offspring Samples

A first-degree BP-I family history is a strong risk factor, as heritability rates range between 40–70% in twin and family studies, and the interaction of family environment, psychosocial stressors and genetic loading also increases BP-I risk³⁷. Furthermore, youth with major depressive disorder (MDD) were significantly more likely to develop BP-I if they had a BP-I family history (50% vs. 10%)³⁸. However, BP-I offspring are more likely than offspring of healthy families to develop any kind of psychiatric disorders, and although mood disorders are even more likely, BP-I development has remained relatively low^39–41. Thus, for youth with genetic loading for BP-I, additional risk factors are needed. Those are also crucial because in community samples, the minority of adults with BP-I has a first degree family member with BP⁴², although prepubertal onset BP-I appears to have the highest familial loading⁴³.

Nevertheless, the genetic enrichment in offspring studies may not only increase the ultimate likelihood of BP-I development in patients with mood symptoms, but this strategy may also enable a more targeted assessment of potentially more specific prodromal symptoms. Only few prospective BP-I offspring studies are available. Compared to offspring of healthy parents, Duffy and colleagues^44,45 in a 15-year follow up study described the early course and onset of illness in those offspring later developing BP-spectrum disorders as generally following a sequence from non-specific psychopathology, followed by minor mood disturbances, depression and finally (hypo-)mania. However, not all individuals progressed through all stages, and BP-I conversion rates have ranged from 3%^44,45 to 11%⁴⁶ during up to 15 years of follow-up. Among non-specific psychopathology, sleep disturbances and anxiety disorders were particularly prevalent, and depression was the most common index episode. Somewhat surprisingly, no pre-pubertal onset (age under 12 years) was found in this sample, and the mean age of onset was 17 years ⁴⁷. While other groups also did not find transitions to mania prior to adolescence in BP-I offspring, but rather reported an elevated rate of a range of psychopathology, mostly anxiety and minor mood disturbances, with a subsequent development of major mood disorders^48–50, a recent, large, retrospective study from 7 international centers (n=1,665) showed clustering of prepubertal-onset illness in familial type BP-I⁴³.

Cyclothymic-hypersensitive temperament

Although adult studies showed some correlation between cyclothymic-hypersensitive temperament (CHT) traits, i.e. a temperament characterized by mood swings and hypersensitivity, and BP-I risk^51–3, results are inconclusive and only few studies investigated temperamental features as risk syndromes for pediatric BP. One prospective study that focused on 80 youths with a current or past clinical diagnosis of depression reported a 2-year conversion rate to bipolar disorder (defined by a two-day threshold for (hypo)mania) of 43%. In this study, youth with CHT, characterized by irritability, aggressive behaviors, anger outbursts, emotional hypersensitivity and mood lability, showed more BP-spectrum conversions than those without CHT (68.8% vs. 15.2%)⁵⁴. Furthermore, those with CHT experienced more frequent and more severe mood episodes of either polarity. Furthermore, patients with CHT and higher likelihood of conversion exhibited significantly higher rates of comorbid conduct disorder (CD) (55.3% vs. 15.2%) and had more frequent suicidal ideation (80.9% vs. 36.4%)⁵⁴.

Unipolar depression

A survey by the National Depressive and Manic-Depressive Association (NDMDA) revealed that MDD was the first sign of BP-I in almost 50% of patients^5,9. Moreover, conversion rates from MDD to BP-I ranged between 7.7–49% in prospective studies, with higher rates among those with preadolescent and adolescent onset MDD and those with longer follow-up^38,55–61. However, reports of conversion from MDD to BP-I below 7% also exist^46,62, and predictors for a differentiation between unipolar and bipolar depression remain inconclusive. One study of 60 hospitalized adolescents followed over 3–4 years found precipitous symptom onset, psychomotor retardation and psychosis to significantly predict conversion to BP-I³⁸. Additional predictive symptoms were depressed mood, self-reproach or self-pity, suicidal tendencies, weight gain, somatic preoccupations, diminished concentration, irritability, and demandingness. Precipitous MDD onset was also predictive of bipolar depression vs. MDD in a study of adults⁶³. Moreover, another study of depressed youth also found that psychotic features and psychomotor retardation, as well as treatment-induced (hypo-) mania and/or a BP-I family history predicted BP-I⁶⁴.

In a follow-up study of a large community sample (n=3,206) in Germany, the incidence of mood disorders was studied in individuals initially aged 14–24 years who were interviewed three times after baseline with a follow-up of 7.3–10.6 years at the last interview⁶⁵. A total of 3.6% of the initial MDD sample (n=649) developed BP-I or BP-II by the 3^rd decade of life. Of note, the risk was highest in adolescent-onset depression before age 17, with a 9% conversion to BP-I/II, without gender differences. Conversely, only 1% of adolescents and young adults who had a minor depressive episode at baseline (n=327) converted to BP-I/II.

Finally, a study of 165 children 8–11 years old with mood disorders followed 37 children with depression-spectrum disorders and transient manic symptoms (DSD+TMS), 13 with DSDs, and 115 with bipolar-spectrum disorders (BSDs) for 18 months⁶⁶. Half of the patients each received Multi-Family Psychoeducational Psychotherapy (MF-PE) after baseline or after a one-year wait-list condition. TMS were defined as: ≥1 DSM-IV mania criterion, except for irritability, that is distinguishable from ADHD, presence of 1–3 manic-like episodes of moderate to severe intensity lasting ≥4 hours or presence of ≥2 brief episodes of 2–4 hours duration, evidence of functional impairment and lack of having a BSD (i.e., BP-I, BP-II, cyclothymia or BP-NOS). At 18 months, conversion to BSD (in patients with follow-up data) was significantly higher for the DSD+TMS group (48.0%, i.e., BP-I=3 (8.1%), BP-II=5 (13.5%), BP-NOS=3 (8.1%), substance-induced mood disorder=1 (2.7%)) compared to the DSD group (12.5%, i.e., BP-NOS: n=1). Poorer baseline functioning and receiving MF-PE after 1 year rather than immediately were associated with BSD conversion (60% vs 16% BSD). These results suggest that, although a careful assessment of sub-threshold mania symptoms in depressed youth is indicated, defined predictors for BP-spectrum disorder risk are lacking. Furthermore, other approaches for early identification are needed as solely focusing on youth with MDD would systematically miss 50% of youth at risk, who develop BP-I without a first MDD episode.

Attenuated (mania) symptoms

The strategy of defining attenuated variants of cardinal/diagnostic symptoms as an at-risk state for developing the full disorder has been fruitfully applied to the early identification of psychosis in samples consisting mainly of young adults^11,12, although approaches are also being applied to adolescents^12,67.

Retrospective studies

Since the clinical high-risk psychosis field developed out of retrospective assessments of the symptomatic prodrome prior to first-episode psychosis, which informed the development of rating instruments and risk criteria that were then prospectively validated¹², a similar approach is also being taken in the emerging BP prodrome field.

To date, still only few studies assessed the initial prodrome to mania. The largest retrospective study of the National Depressive and Manic-Depressive Association (n=500)⁹ revealed that adults with a self-reported diagnosis of BP-I experienced a variety of prodromal symptoms. The most common symptoms included depressed mood or hopelessness (33%), mania/hyperactivity (32%), decrease in sleep (24%), and mood swings (13%). As this study surveyed adults, it is unclear if these differ from prodromal symptoms in children and/or adolescents. However, a chart review of Amish youth who developed BP-I also lists mania and depression-related symptoms, including depressed mood (53%), increased energy (47%) and decreased sleep (26%), as commonly reported prodromal symptoms in youth⁶⁸. However, while adults reported non-specific symptomatology (i.e., anger, irritability, psychotic symptoms) to a lesser degree (9%), these symptoms were a more significant part of the prodrome in youth (anger (38%), irritable mood (33%), bold/intrusive/excessive behaviors or conduct difficulties (28–29%), crying spells (26%), being overly sensitive (24%)). Another study⁶⁹ that asked parents via mailed questionnaires about symptoms during each year of their child’s life, who had either a community diagnosis of BP-I (n=78), psychiatric diagnosis other than BP-I (n=38), or no psychiatric diagnoses (n=82), extended the results above, reporting that irritability/dyscontrol were the earliest features that distinguished children that later developed BP-I from those who did not, while depressive and manic symptoms also distinguished the groups, but only in later childhood, i.e., at age 7–8 years. Psychosis and suicidality were the last symptoms that distinguished the diagnostic groups at around age 9. Likewise, Correll and colleagues⁷⁰ assessed prodromal symptoms in youth with recent-onset BP-I or BP-II using a structured interview (Bipolar Prodrome Symptoms Scale-Retrospective, BPSS-R), capturing detailed information on the timing, frequency and severity of symptoms. They also found a high proportion of depressive and mania-related symptoms (depressed mood (54%), anhedonia (40%), insomnia (36%), feelings of worthlessness (33%), irritability (61%), increased energy (50%) and psychomotor agitation (48%)). Furthermore, again non-specific symptoms were prominent, including decreased functioning (65%), mood lability (58%), anger outbursts (48%), social isolation (44%) and anxiety (42%). In this study, the, mean prodromal phase before mania onset lasted 18 months, a time span allowing early identification efforts.

Additionally, Ritter and colleagues⁷¹ point to the significance of sleep disturbances for mood disorders in general, and for BP-spectrum disorders in particular, concluding that sleep disturbances did not begin in childhood, but emerged in adolescence, being a significant risk factor for BP-spectrum disorders.

Prospective Studies

Following a large cohort of college students for over 8 years, Lewinsohn et al.⁷² studied a subgroup of youth (mean baseline age 16.8, n=48) with “subsyndromal bipolar” disorder defined by “distinct period of abnormally and persistent elevated, expansive, or irritable mood, plus >1 manic symptom”, finding a very low conversion to BP (1/48=2.1%). Conversely, 40.9% of these youth developed MDD. However, in an analysis of the Munich sample (n=3,206) mentioned earlier⁶⁵, the predictive value of the number and persistence of mania-like symptoms for BP-I or II development was assessed prospectively in a risk sample of 1,565 subjects aged 14–21 at baseline and followed for 10 years⁷³. The authors found a significant dose-response relationship for persistence and number of both manic and depressive symptoms. Moreover, in 30–40% of individuals converting to BP-I/II the symptom progression could be traced back to subthreshold manic and depressive symptoms⁷³.

As reviewed above, only one of the cited studies investigating the bipolar prodrome assessed non-bipolar comparison groups regarding childhood symptoms. Including control groups is crucial, as prodromal symptoms may overlap between different disorders. This was also shown by a small case series of 47 and 26 adolescents and young adults who either fulfilled ultra high-risk criteria for psychosis⁷⁴ or met criteria for psychosis NOS⁷⁵, but developed BP-I in 6.5% and 15.4% (19.2% when counting in BP-NOS), respectively, with anxiety symptoms distinguishing patients developing BP-spectrum disorders from those developing non-affective psychoses⁷⁶. Finally, in a prospective study of adolescents fulfilling clinical high-risk criteria (i.e., attenuated positive or negative symptoms) for schizophrenia-spectrum psychoses, 8 adolescents who eventually developed BP-I (n=6, 5 with psychotic features) or BP-NOS (n=2) were compared on baseline symptom and neurocognitive measures to 24 adolescents who developed schizophrenia-spectrum psychotic disorder and 115 non-converters⁷⁷. In this study, BP and schizophrenia-spectrum outcome patients differed from non-converters, but not from each other, highlighting the need for BP-specific assessments and instruments. In addition, it is important to find predictors that distinguish between the development of affective and non-affective psychotic disorders.

Two studies used scores on the General Behavior Inventory (GBI)^78–80 to symptomatically define BP risk in youth. In a 5-year follow-up study, higher initial depression scores on the GBI differentiated individuals who later developed bipolar and other mood disorders from those without mood disorders, and those youth who converted from depression to BP-spectrum disorders from those who retained a depression diagnosis⁸¹. However, no validated cut-off scores are reported for the GBI yet to estimate the risk for BP-spectrum disorders in clinical practice. Furthermore, in the ongoing Longitudinal Assessment of Mania Study (LAMS)⁸², children 6–12 years old are prospectively evaluated regarding correlates and the predictive utility of “elevated mania symptoms” (EMS) defined by a tentative cutoff score on the GBI of ≥ 12, for future BP development. Preliminary results of 707 subjects showed that children with EMS had significantly more BP-spectrum (BP-I: 10.6% vs. 5.8%, BP-II: 0.5% vs. 0%, cyclothymia: 1.8% vs. 0%, BP-NOS: 12.1% vs. 2.3%), ADHD and disruptive behavior disorder diagnoses than children with lower GBI scores, without differences in depressive disorders⁸³. However, conversion rates to BP are lacking.

Furthermore, a certain profile of attention, aggression and anxiety/depression symptoms on the Child Behavior Checklist (CBCL)⁸⁴ has been discussed as a pediatric BP phenotype after a meta-analytic review of seven studies⁸⁵. However, despite some supportive findings,^86–89 the majority of subsequent cross-sectional and longitudinal studies^90–96 did not support the CBCL or the pediatric BP subscale sum score as standardized measures enabling the differentiation of BP from other psychopathology ().

Bipolar–spectrum disorders

BP-NOS can be conceptualized as the most distal prodrome to BP-I (unless one counts cyclothymia and BP-II as prodromal states to BP-I), as many, but not all symptomatic criteria of BP-I are fulfilled. In a landmark study⁹⁷ that prospectively studied 92 youth (mean age: 12.1 years) with clearly operationalized BP-NOS, conversion rates after 2.5 and 4 years, were 20% each to BP-I and 10% and 18% to BP-II. In this study, BP-NOS was defined by either one less B criterion in the presence of irritability or expansive mood than required by DSM-IV, or a minimum of 4 days of at least 4 hours of suprathreshold criteria present cumulatively, rather than consecutively, plus functional impairment. However, since more than 80% of youth fulfilled all DSM-IV mania criteria at baseline, except for the duration criterion, and as in younger subjects ultradian cycling and very brief episodes are common, it is still unclear if these youth biologically truly had subsyndromal mania or if they had a potential pediatric mania subtype with shorter episode duration which morphed into a more adulthood pattern in those patients who “transitioned” to mania, as they biologically matured, expressing longer mood episodes. To date, conversion rates and patterns in youth or adults diagnosed with mood disorder NOS have not been reported.

Conceptual Models of the Bipolar Prodrome

Several groups proposed staging models and preliminary criteria for the BP prodrome (Table 2). McNamara and colleagues’¹⁹ model entails three stages for the development of BP-I, i) putative risk factors (including a first-degree relative with BP-I, sexual/physical abuse, psychosocial stress, substance use/abuse, stimulant/antidepressant exposure, n-3 fatty acid deficiency), ii) “prodromal” clinical features (ADHD, MDD, hypomania, anger/irritability, anxiety), and iii) mania/bipolar disorder-I onset. The authors suggest staged treatments according to the increasing risk, i.e., benign interventions as omega-3-fatty acids, vitamins and psychotherapy for stage 1, stimulants and mood stabilizers for stage 2 and potentially antipsychotics for stage 3.

We have proposed¹⁵ a symptom-based BP I and II staging model with a progression from an early prodromal stage - consisting of sub-clinical, non-specific and attenuated mania symptoms - to a late prodromal stage – defined as BP NOS - and further to a subsyndromal (cylothymia, hypomania) and syndromal (BP-I and II) stage. As suggested before¹⁵, these stages should be studied concurrently with a clinical, endophenotypic, and genetic approach.

Based on data from the Australian early identification program, Bechdolf and colleagues¹⁸ proposed three prodromal criteria groups for BP-I: 1) sub-threshold mania, 2) depression plus cyclothymic features, and 3) depression plus genetic risk. A preliminary assessment of conversion rates in 173 patients aged 15–25 years categorized at-risk for BP by these criteria in retrospective chart reviews showing significantly more conversions (22.7% vs. 0.7%, p<.001, defined as hypomania/mania-related initiation or change in treatment) after 265.5±214.7 days in the group of at-risk subjects (n=22) than the non-risk subjects (n=151)¹⁸.

Leopold and colleagues²³ defined six domains of risk criteria for BP-I that they suggest for prospective investigation: i) family history, ii) impairment in psychosocial functioning, iii) pronounced creativity, iv) ADHD symptoms or diagnosis, v) substance use/abuse, vi) changes in sleep and circadian rhythm, mood swing and lability, anxiety, depressive, hypomanic or mixed symptoms, and dissociative symptoms. Similarly, Brietzke and colleagues⁹⁸ defined three risk clusters for BP-I, i.e., a genetic risk, environmental risk and a profile of risk biomarkers.

As proposed before¹⁵, Pavuluri⁹⁹ suggested that investigating endophenotypes, which could differentiate between childhood diagnoses that often precede BP (i.e., ADHD and prepubertal MDD) and early, prodromal manifestations of BP enhances early identification strategies. In addition, markers of emotional, social and cognitive dysregulation family history and other neurobiological and temperamental markers should be studied.

Conclusions and Implications for Research and Clinical Practice

The search for symptoms and factors that can reliably predict BP-I development in youth is in its early stages. To date, several strategies point to phenomenological candidates that serve as the basis for the recent development of structured assessment tools and criteria for the BP-I prodrome^{18,23,70,79–83} that will have to prove their predictive value. Prospective research will have to continue the refinement and definition of potential risk markers and clinical predictors. With the parallel advance of knowledge on BP-I endophenotypes in youth, including genetics, neuroimaging, neurochemistry and neurocognition, a combination of both, clinical and biological factors, is expected to enhance pre-syndromal identification of BP-I¹⁵. Since the main age of onset of the first illness signs appears to span adolescence and early adulthood, close collaboration between or, ideally, integration of child and adolescent psychiatric services with adult psychiatric services should be attempted in the research and optimal care of at-risk youth. Furthermore, since identified youth with clinical risk status for BP-I will include patients at true risk for other psychiatric disorders and outcomes, a more inclusive research approach covering the assessment and treatment of a broader range of emerging psychopathology should be considered.

Until a more precise risk prediction can be achieved, clinicians should carefully assess current putative risk symptoms for BP-I (i.e., depressive and attenuated mania-like symptoms, CHT traits, decline in social and role functioning, ADHD and DBD) in patients with mood-spectrum presentations (Table 3), particularly in youth with major depression or a BP-spectrum family history. This information should be used to aid clinical decision making regarding the early diagnosis and treatment. In youth at risk for BP, treatment options should be weighed carefully and antidepressants may need to be avoided in favor of mood stabilizer treatment with antidepressant activity, or surveillance be increased if antidepressants are started. In general, as proposed for psychotic disorders¹⁴, a staged treatment approach to BP-I risk constellations that uses benign, nonpharmacologic interventions first and increases the treatment-related risk as the risk and severity of the suspected or diagnosed conditions increase²⁵, is considered the most appropriate approach, especially in youth who are even more sensitive to medication adverse effects^100,101.

Table 3.

Different Concepts of the Bipolar-I Disorder Prodrome

Correll et al. 2007 15	McNamara et al 2010 19	Bechdolf et al. 2010 18	Pavuluri 2010 86	Leopold et al. 2012 23	Brietzke et al. 2012 85
4 illness stages: early prodromal phase: sub-clinical and non-specific symptoms, attenuated mania symptoms late prodromal phase: bipolar disorder NOS subsyndromal stage: cyclothymia, hypomania syndromal stage: bipolar disorder-I and II 3 approaches: clinical: (e.g. attenuated symptoms, ADHD, ODD, depression etc.) endophenotypic genetic (including familial risk and genotype)	3 illness stages: putative risk factors: first-degree relative with bipolar disorder, sexual/physical abuse, psychosocial stress, substance use/abuse, stimulant/antidepressant medication exposure, n-3 fatty acid deficiency prodromal clinical features: ADHD, unipolar depression, hypomania, anger/irritability, anxiety mania/bipolar-I disorder onset	3 risk groups: sub-threshold mania group presence of an A criterion plus at least two B criteria, duration of 2–4 days depression plus cyclothymic features group depression: depressed mood or loss of interest plus 2 depression criteria, duration ≥1 week cyclothymic features: numerous episodes with sub-threshold mania as in group I and depression, duration of sub-threshold mania 4hrs/day for ≥4 cumulative days lifetime depression plus genetic risk group depression as in group II and first-degree relative with bipolar disorder	5 marker domains: family history ‘prodromal symptom complex’ including symptoms of mania, comorbid ADHD and/or ODD, and temperamental features neurobiological markers endophenotypic marker temperamental features	3 risk stages: risk state specific changes in sleep and circadian rhythm plus ≥1 other secondary risk factor high risk state one primary plus ≥1 secondary risk factor ultra-high risk state >1 primary risk factor Primary risk factors: family history increasing cyclothymia dynamic (hypo-)mania prodrome Secondary risk factors: changes in sleep impairment in psychosocial functioning symptoms or diagnosis of ADHD substance use/abuse anxiety current or past mood disorder other than bipolar disorder	3 risk clusters: genetic risk environmental risk childhood trauma, sexual/physical abuse, neglect or parental loss profile of risk biomarkers

ADHD: attention deficit-hyperactivity disorder; nos: not otherwise specified; ODD: oppositional defiant disorder

Clinical Implications.

1. A clinical prodrome preceding a first episode of mania seems to exist in a majority of patients developing bipolar I disorder (BP-I), including attenuated mania-like symptoms, depressive symptoms as well as non-specific global illness signs, such as decline in social and academic/role functioning. However, no sufficiently predictive risk constellations have been identified so far that could be applied in clinical practice.

2. Family history of BP-I and presence of syndromal depression, attention deficit-hyperactivity disorder, disruptive behavior disorders, and cyclothymic hypersensitive temperamental traits are risk factors for future BP-I development, but most subjects exhibiting these risk factors do not seem to develop BP-I. This increases the need for the identification of genetic, neuroimaging, neurochemistry and neurocognition endophenotypes that will further enhance early identification of BP-I in subsyndromal phases in the future.

3. Clinicians should carefully assess potential BP-I risk symptoms in patients with mood-spectrum presentations, particularly in youth with major depression or a BP-I family history, carefully weighing and monitoring the potential risks and benefits of antidepressant treatment.

Limitations.

1. Research on the prodrome to BP-I is still in its early stages, and phenomenological description, risk marker identification, and development of specific rating tools and operationalized criteria for the BP-I/mania prodrome are still preliminary.

2. Appropriate target groups and enrichment strategies for prospective studies have yet to be identified, and the specificity of the prodrome has to be determined in prospective studies that include appropriate psychiatric control groups.

3. A multi-stage treatment model, ranging from low risk, non-pharmacologic interventions to interventions with higher treatment-related risks according to an increasing risk of BP-I, still needs to be defined.