Chronic Irritability in Youth: A Reprise on Challenges and Opportunities Toward Meeting Unmet Clinical Needs

INTRODUCTION

Emotion dysregulation is one of, if not the, most common reason that youth are brought to psychiatric care. The impact of emotion dysregulation is experienced, not only by youth, but also as by their families, as described in the parent perspective by Pavlov-Blum that opens volume I.¹ This commentary adopts a translational perspective on one subtype of emotion dysregulation, irritability. A translational perspective is key to elucidating pathophysiological mechanisms of irritability; elucidating such mechanisms is, in turn, central to the development of novel, personalized interventions.

We first focus on challenges and an opportunity that arise when studying irritability from a translational perspective. One translational challenge concerns defining the construct of interest.

For example, in a volume on emotion dysregulation, why focus specifically on irritability? Our discussion here complements the work presented by Connor and Doerfler and Althoff and Ametti.¹ We “drill down” into the construct of irritability; our working definition of irritability is “elevated proneness to anger, relative to peers.”² We discuss external validity of irritability and differing, but closely associated, variations in its presentation (ie, phasic vs tonic irritability). We then discuss the related challenge that, at least in clinical populations, irritability rarely occurs alone. In youth selected specifically for severe irritability (ie, those with disruptive mood dysregulation disorder [DMDD]), the most common comorbid diagnoses are anxiety disorders and attention-deficit hyperactivity disorder (ADHD). It is important to identify shared versus unique mechanisms among irritability, anxiety, and ADHD in order to predict the course of children presenting with complex combinations of these phenotypes and guide the development of personalized treatments. Here, our discussion complements the Blader, Walkup and colleagues, and Weisbrot and Carlson articles.¹

We then turn to basic mechanisms afforded by a focus on irritability, namely the utility of the construct of frustrative nonreward (FNR) in studying the neural mechanisms of severe irritability in youth. FNR is a construct and behavioral paradigm developed first in rodents and studied in nonhuman primates and humans.³ In a heuristic model of irritability published in 2017,² we posited that irritability can be understood as aberrant responses to frustration and to threat. We focus specifically on studies of frustration and include a discussion of the implications of this work for the development of novel interventions.

TRANSLATIONAL CHALLENGES IN STUDYING IRRITABILITY

Defining Irritability and Overlapping Constructs

Multiple constructs and definitions have been used to capture the clinically important severe dysregulation that is the focus of this special edition (eg, Spring and Carlson, Keluskar and colleagues, Ashurova and colleagues, Keeshin and colleagues, Benton and colleagues, and Sharma and McClellan).¹ See Fig. 1 for one conceptualization of some of these constructs. Dysregulation itself, sometimes operationalized using the Children’s Behavior Checklist (CBCL) Dysregulation Profile, is one construct, encompassing dysfunction across emotional, behavioral, and attentional domains (see Connor and Doerfler and Althoff and Ametti, volume I).¹ Lability and irritability are narrower, as their definition is based on emotion and mood, with associated behavior seen as secondary. Lability can be defined as abnormally frequent shifts into both negative and positive emotions and moods. It features prominently in the literature on bipolar disorder (BD), where euphoric mood is associated with hedonic behaviors. Also, the clinical observation that youth with ADHD may have uncontrollable bouts of “silliness” sparked interest in emotional lability in ADHD and was sometimes viewed as supporting a diagnosis of mania in these youth.^4–6

Fig. 1.

Select overlapping constructs used to describe emotional and behavioral dysregulation. The figure highlights the increasing specificity and narrow scope of the constructs. See text for definition and discussion of each construct.

Irritability has had multiple definitions and conceptualizations; for a thorough review, see Toohey and DiGiuseppe.⁷ One parsimonious conceptualization specifies that, like lability, irritability is characterized by frequent and marked shifts of mood and emotion, but in irritability the shifts are only into a negative mood state, specifically anger. Behavioral outputs of irritability can then include explosive temper outbursts and reactive aggression (see Spring and Carlson and Connor and Doerfler, volume I).¹

Our group’s working definition of irritability is “elevated propensity to anger, relative to peers.”² This definition is admittedly imperfect and likely to require modification as we learn more. The definition was influenced by the conceptualization in Fig. 1; our aim was to be narrow and specific enough to enable us to frame tractable research questions while being broad enough to encompass important clinical phenomena. Importantly, because our research aims at identifying brain mechanisms of irritability, we wanted a construct that allows for translational bridges, as this does in linking to anger, aggression, frustration, and threat.² We wanted a dimensional construct that encompasses psychopathology. Our definition does not allow for “normative irritability,” which can be seen as a limitation, nor does it specify a threshold for “elevated.” Also, arguably the definition conflates mood and emotion, and its conceptualization of behavior as secondary to mood and emotion needs to be confirmed by further research.

External Validity of the Irritability Construct

How useful is the irritability construct in furthering clinically meaningful research? To assess this, we use the framework for “external validity” developed Dr Kenneth Kendler8 for the Diagnostic and Statistical Manual of Mental Disorders, Fifth Revision (DSM-5) process. External validity was defined as reliably providing relevant clinical information. Six high-priority validators were defined: (1) familial aggregation and/or co-aggregation (eg, family, twin, or adoption studies); (2) biological markers (eg, molecular genetics, neural substrates); (3) degree or nature of functional impairment; (4) diagnostic stability; (5) course of illness; and (6) response to treatment. Although these criteria were designed to assess categorical diagnostic constructs, here we apply them to the dimensional construct of irritability. The question is whether knowing if an individual is relatively high or low on trait irritability provides useful cross-sectional and predictive information. Because, in clinical samples, irritability rarely occurs alone (see later in this article), the goal of these studies is to identify associations with irritability specifically, rather than with comorbid traits or diagnoses.

First, in terms of familial aggregation, family studies show a link between parental (especially maternal) depression and child irritability, with the causal arrows usually being bidirectional.^9-11 Perhaps more compellingly, twin studies show the heritability of irritability to be in the moderate range (ie, approximately 0.3–0.6), similar to anxiety and unipolar depressive disorders.^12-17 Thus, there appears to be good evidence for familial aggregation.

Second, regarding biological markers, for irritability as for all other psychological traits and psychiatric diagnoses, the search for biomarkers remains ongoing and aspirational; hopefully the research described later in this article and other ongoing work further this effort.

Third, there is good evidence that irritability can be impairing in multiple domains (home, school, peers).^18,19 Further, comorbid irritability is associated with increased impairment in youth with illnesses such as anxiety or ADHD.^20,21 Youth irritability appears to be associated with impairment into early adulthood, again after controlling for other symptoms.²²

The fourth criterion for external validity, stability, was articulated for DSM-5 in terms of stability of categorical diagnoses; here we focus primarily on evidence for stability of the dimensional trait. In a large, community-based twin sample (N = 1431 pairs) with 1-year follow-up, there was moderate stability for both the categorical and dimensional formulations of DMDD/irritability.¹⁵ However, other evidence suggests that dimensionally operationalized trait irritability may be more stable than categorical DMDD. For example, in a 2-year to 4-year follow-up of 200 youth with severe mood dysregulation (SMD, the precursor phenotype to DMDD²³) in community treatment, the % meeting SMD criteria declined from 100% at baseline (mean age 11.6 ± 2.6 years) to 49% at 2 years and 40% at 4 years, indicating poor stability.²⁴ However, throughout the follow-up period, youth remained moderately impaired, with markedly stable Clinical Global Impression (CGI) SMD and CGI overall impairment ratings. Notably, more than 80% continued to receive psychotropic medications and more than 20% had been hospitalized. Major limitations of this study include poor retention, possibly introducing bias. Similar results were found in a smaller (N = 36) and younger (ages 6–9 years) sample.²⁵ Of note, however, some studies do not find good stability, even for dimensionally assessed irritability.^26,27

The decreased stability of categorical versus dimensional approaches to irritability may reflect arbitrary and nondevelopmentally sensitive thresholds for the category. Indeed, in a community-based sample of preschoolers (N = 425, mean age 5 4.7 years), dimensional measures were used to derive a severity cutoff to identify potentially concerning irritability. A 2-item measure identified diagnostic status at baseline and persistent irritability, diagnoses, and impairment at follow-up (mean = 2.9 years), showing good stability.¹⁹

Studies that use latent class longitudinal approaches to differentiate trajectories of irritability can provide more detailed information about stability. These studies identify classes of youth that differ in the developmental course of their irritability. The most common class (ie, the normative trajectory) shows decreasing irritability from preschool through adolescence.^28–30 Classes can be compared on clinical as well as neural features,²⁹ and can be useful in future studies of staging and prognostication. Classes of particular interest to clinicians include youth with persistently moderate to high irritability.

The fifth external validity criterion is course of illness. Applying this criterion to a dimensional trait, one can ask whether irritability is associated with an increased risk for the emergence of specific symptoms or diagnoses over time. As noted previously, longitudinal research conducted in the context of questions about irritability and pediatric BD demonstrated an association between childhood irritability and increased risk for unipolar depression and anxiety disorders (for a meta-analysis, see Vidal-Ribas and colleagues³¹). Twin studies suggest that this longitudinal association is, in part, genetically mediated.^12–17 In addition, multiple studies show longitudinal associations between the irritable dimension of oppositional defiant disorder and risk for unipolar depression and anxiety (see Burke, volume II).¹ However, the literature is not entirely consistent, as studies that differ in methodology and sampling find more protean longitudinal associations between childhood irritability and future psychopathology.^32,33 Future research parsing the heterogeneity of irritability (eg, threat-based vs reward-based irritability and tonic vs phasic irritability, see later in this article) may allow for more personalized and accurate predictions. It is also important to note longitudinal associations between earlier irritability and later suicidality (see Benton and colleagues, volume I),^1,22,30,34 and new data showing associations between irritability and suicidality in depressed adults.³⁵

Finally, does the presence of irritability predict treatment response? Few studies specify irritability as an outcome, and adequate measurement of irritability is perhaps the most important area for future research to facilitate the development of novel treatments (see Carlson and Singh, volume I/II introductions, and Wozniak and colleagues, Patino and DelBello, Volume II).¹ A post hoc analysis of the MTA study on the treatment of ADHD found that methylphenidate (MPH) decreased irritability in youth with ADHD.³⁶ Although this effect was not attributable to the effect of MPH on ADHD symptoms, it is unknown whether stimulant medication would reduce irritability in youth without ADHD. As mentioned previously, a trial of MPH plus citalopram versus MPH versus placebo showed promising results,³⁷ and again the effect of treatment on irritability was not secondary to effects on anxiety or depression. Further, in adults with major depressive disorder (MDD), decreased irritability early in SSRI treatment was a favorable predictor of remission³⁵ and data suggest that, in depressed adults, sertraline decreases irritability independent of its effect on depression.³⁸ Thus, there is emerging evidence that irritability itself may respond to specific treatment interventions.

There is also evidence for specific psychotherapeutic interventions in school (Mattison and colleagues, volume II), inpatient (Chua and colleagues, volume II), residential (Huefner and colleagues, volume II), and outpatient (Waxmonsky and colleagues, Singh and colleagues, Fristad and colleagues, volume II) settings.¹ A recent analysis of youth (N = 81) with severe irritability found that those treated with Modular Approach to Therapy for Children with Anxiety, Depression, Traumatic Stress, or Conduct Problems (MATCH), a modular, transdiagnostic, behavioral/cognitive-behavioral intervention, had faster improvement than those who received standard manualized treatments (SMT) or usual care (UC). SMT and MATCH both surpassed UC in irritability reduction (effect size 5 0.49) (see Evans and colleagues, volume II).¹ A trial of dialectical behavior therapy in pre-adolescents (N = 43) also showed feasibility and preliminary efficacy in youth with DMDD.³⁹

In summary, how would we assess the external validity of irritability? The data provide at least moderate support for familial aggregation and impairment. There is modest to moderate evidence for stability of the dimensional trait and the ability of irritability to predict future clinical course. Given the paucity of research, there are only suggestive data for specific treatment responses, and currently there is no clear evidence for specific biological markers. We conclude that this qualitative analysis suggests sufficient external validity for the construct to justify future research, although quantitative, systematic assessment is warranted. From the research and clinical perspective, we suggest that priority be placed on longitudinal designs to ascertain stability and prediction of clinical course, systematic treatment trials, and discovering biological markers that might facilitate both diagnosis and the development of novel interventions.

Tonic versus Phasic Irritability

Although irritability is a narrower construct than some others in Fig. 1, it is nonetheless complex. As discussed in Spring and Carlson (volume I),¹ the importance of distinguishing chronic irritability from irritability occurring in the context of a depressive or manic mood episode contributed to confusion regarding the diagnosis of pediatric BD. However, even within the context of chronic irritability, there is phenomenological variation, including in the form of tonic and phasic irritability.

Our first attempt to operationalize irritability occurred when we specified criteria for the SMD phenotype.²³ These served as inclusion criteria for studies differentiating youth with “classic,” episodic BD from those with the controversial phenotype of chronic severe irritability, operationalized as SMD. In defining SMD, we specified 2 types of irritability: phasic (ie, developmentally inappropriate temper outbursts) and tonic (ie, chronically grumpy, angry mood between outbursts). This differentiation was done on clinical grounds, with the observation that some children (anecdotally, often children with ADHD with comorbid oppositional defiant disorder) have normal mood between outbursts. In studies contrasting chronically irritable youth to those with clear, episodic BD, we wanted to limit the chronically irritable group to youth who had, not only extreme and frequent temper outbursts, but also a persistent mood disorder between outbursts. This chronically irritable mood contributed to their receiving the diagnosis of BD and to severe impairment, making them a more suitable match for the youth with “classic” BD.

A major question is whether tonic and phasic irritability are distinct constructs with differential mechanisms and predictions (see Vidal-Ribas and Stringaris and Spring and Carlson, volume I, and Burke and colleagues, volume II).¹ Indeed, in assessing the 2 constructs, tonic irritability is frequently assessed by asking about the ease of precipitating an outburst, so in practice it can be difficult to rate the 2 constructs independently. Parenthetically, the DMDD criteria for tonic irritability is modeled on the “A” criterion for MDD; both require that the abnormal mood be present most of the day, nearly every day, and these 2 diagnoses both received “questionable” kappas in the DSM-5 field trials.⁴⁰ Whether this low reliability is due to specific challenges associated with assessing chronic mood abnormalities is a question deserving of further research.

In both clinical and community samples, tonic and phasic irritability are highly correlated. In a sample of 489 youth including healthy subjects, DMDD youth, ADHD youth, and youth with irritability symptoms not meeting the DMDD threshold, the correlation between phasic and tonic irritability was 0.79 (Cardinale and colleagues, in press).⁴¹ In the Great Smoky Mountain community sample (N = 1420, ages 9–16 years), 51.4% of subjects reported phasic irritability (ie, a tantrum at least once in the previous 3 months), 28.3% tonic irritability, and 22.8% both.⁴² Overlap was high, and tonic irritability rarely occurred without phasic irritability. Thus, research attempting to dissociate the two requires large samples with adequate representation in the discordant cells.

Nonetheless, there is preliminary genetic, comorbidity, and treatment evidence supporting differentiation of the 2 constructs. In each instance, phasic irritability appears to carry stronger predictions than does tonic irritability, perhaps reflecting greater reliability in assessment. Regarding genetic evidence, a study of 1431 twin pairs, aged 8 to 17, found that, compared with tonic irritability, phasic irritability was somewhat more stable and heritable and contributed more to a latent variable designed to capture the DMDD phenotype.¹⁵ Further, only half the genetic variance of tonic irritability was shared with phasic irritability. Regarding comorbidity, in the sample of 489 youth oversampled for irritability and ADHD described previously, high comorbid ADHD symptoms were associated with greater phasic irritability than were moderate levels of ADHD symptoms, whereas there was no differences between groups in tonic symptoms (Cardinale and colleagues, in press).⁴¹ These results complement recent work on an “irritable subtype” of ADHD by focusing attention on youth with ADHD and phasic irritability.⁴³ Finally, regarding treatment, in the open stimulant-only phase of the citalopram 1 MPH trial described previously, MPH reduced temper outbursts and ADHD symptoms, but not irritable mood between outbursts.³⁷ Thus, there is some preliminary evidence that phasic and tonic irritability may differ in treatment response.

Preliminary ecological momentary assessment (EMA) data suggest that the use of mobile technology may advance research on the differentiation of phasic versus tonic irritability while also clarifying informant effects that impact on the assessment of irritability and other childhood psychopathology.⁴⁴ In EMA, parents and children are prompted simultaneously several times a day on their smart-phones to answer questions about the child’s mood, behavior, sleep, and other relevant variables. This acquisition of real time, contemporaneous ratings from parent and child holds the promise of circumventing the well-known biases inherent in retrospective report and of clarifying informant effects. Preliminary data in 62 youth who sought treatment for irritability (eg, DMDD or sub-DMDD) and/or ADHD indicate (1) an informant effect whereby parents report more symptoms than do youth; (2) evidence that child and parent data are independently informative in predicting temper outbursts; (3) good agreement between clinician characterization of child’s irritability (based on retrospective parent and child report) and EMA measures; and (4) evidence that tonic and phasic irritability are dissociable dimensions (Chue and colleagues and Naim and colleagues, unpublished, 2020). Ultimately, self- and parent-reported EMA data can be paired with measures of the child’s physiology (eg, heart rate variability, skin conductance response, actigraphy), also acquired through mobile technology, with the goal of predicting temper outbursts and, when an impending outburst is detected, delivering to the child’s smartphone prompts to engage in a therapeutic intervention. Smart phone and sensor measurements may be particularly useful in the development of novel primary and secondary preventive interventions as reviewed by Singh and colleagues in volume II.¹

Further research on possible distinctions between tonic and phasic irritability could facilitate more personalized medication approaches and prognoses. In addition, such research could have important pathophysiological and theoretic implications. That is, the phasic irritability construct can be conceptualized as an evoked response to a motivationally important stimulus (eg, a frustrating or threatening event), and is thus consistent with definitions of emotion in the affective neuroscience literature. In contrast, tonic irritability is a longer-duration mood, less tightly tied to a specific precipitant. Also, the most salient features of phasic irritability are the behaviors associated with a temper outburst, although it is also associated with angry emotion. Tonic irritability, in contrast, is defined in terms of mood, although one would also expect to see associated behaviors. Identifying biological hallmarks of these different constructs is important for future work on irritability.

Common Comorbidities: Anxiety and Attention-Deficit Hyperactivity Disorder

Clinicians know that irritability rarely presents alone; the articles in Volume I discuss in detail the many childhood psychopathologies that co-occur with irritability. We focus on 2 of the most common co-occurring diagnoses: anxiety disorders and ADHD (see Blader and also Walkup and colleagues, volume I).¹ In a recent sample of 191 youth with DMDD that we recruited, 72% met criteria for ADHD and 46% met criteria for an anxiety disorder (Cardinale and colleagues, unpublished data). As noted previously, in research designed to elucidate the pathophysiology of irritability, it is important to use recruitment and analytical strategies designed to identify common versus unique pathophysiological mechanisms for example, dimensional approaches in transdiagnostic samples enriched for the phenotypes of interest. Also, because comorbidity is the rule rather than the exception in psychiatry, there is now considerable interest in statistical approaches that can dissociate such mechanisms, for example, bifactor modeling, network analyses, and latent profile analyses.⁴⁵

Anxiety and irritability share psychological commonalities, that is, both can be viewed as high-arousal, negative valence responses to threat.² They differ, however, in the nature of the response, that is, avoidance for anxiety and approach (eg, increased motoric activity, yelling) for irritability. Both anxious and irritable youth show an attention bias toward threat.^46,47 In a functional MRI (fMRI) study of attention bias to threat in youth with anxiety, DMDD, ADHD, or no diagnosis (N = 197), we used bifactor modeling to differentiate neural activity associated specifically with parent-rated irritability, child-rated irritability, or parent-rated and child-rated anxiety, or with a shared common factor of negative affectivity.⁴⁸ (Notably, although parent-rated and child-rated anxiety loaded on one factor, parent-rated and child-rated irritability did not, indicating a marked informant effect.) The major associations were between neural activity and parent-related irritability and between neural connectivity and anxiety. Specifically, parent-reported irritability was associated with abnormally increased amygdala, insula, and dorsolateral prefrontal cortex activity when subjects were asked to focus their attention away from threat. In studies that do not consider irritability, abnormal activity in these regions has been reported in association with anxiety,^49–51 demonstrating the importance of considering both traits simultaneously.

Considerable research has focused on emotion dysregulation in ADHD. Mirroring the previous discussion of overlapping constructs, emotionality has been operationalized in multiple ways in the ADHD literature, including deficient emotional self-regulation, emotional impulsiveness, and irritability.⁵ The question of whether such emotion dysregulation should be conceptualized as a core construct of ADHD, a comorbidity, or characteristic of an ADHD subtype has been the subject of considerable theoretic and empirical work (see Blader, volume I).^1,5,43,52

In this context, it is important to note literature emerging on associations between genetic risk for ADHD and emotional dysregulation. As noted previously, several twin studies have found genetic associations among unipolar depression, anxiety, and irritability. In a complementary literature, recent studies in samples enriched for ADHD use polygenic risk scores (PRS) to investigate genetic associations among irritability, ADHD, and MDD. PRS quantify genetic risk based on the presence of single nucleotide polymorphisms associated with an illness or trait.⁵³

Two recent studies have provocative findings. Riglin and colleagues⁵⁴ used PRS and longitudinal trajectories of irritability to identify 2 irritability subtypes in a community sample (N = 7924). The first subtype, early-onset irritability, was predominately male and associated with childhood ADHD and ADHD PRS. The second subtype, adolescent-onset irritability, was predominately female and associated with adolescent depression and PRS for MDD. This highlights the importance of considering possible sex, as well as age, effects in studies of irritability. Second, Nigg and colleagues⁵ studied ADHD and MDD PRS in a sample of 514 patients with ADHD and without ADHD. This study conceptualized emotion dysregulation in the context of temperament, specifically irritability and surgency, with the latter being a temperament defined as positive valence emotion dysregulation (eg, impulsivity, sensation-seeking). They found that ADHD PRS score was associated with both irritability and surgency. This finding would suggest that ADHD PRS is associated with lability, shifts into both positive and negative valence states, rather than simply the negative valence states characterized by irritability (see Fig. 1). More detailed phenotyping of these mood states would be required to identify their precise time course that is, whether positive and negative mood states occur concurrently or simultaneously (as in mixed states), and the role of tonic versus phasic irritability. In any case, these early studies, and the ones above of anxiety and irritability, demonstrate the importance of identifying unique versus shared mechanisms between irritability and other co-occurring illnesses and traits. We can expect to see considerably more research in this important and emerging area.^46–48

Irritability and the Diagnostic and Statistical Manual of Mental Disorders, Fifth Revision

As described in Volume I, major changes were introduced in DSM-5 in response to increased rates of youth receiving the diagnosis of BD, coupled with data suggesting that this increase might be attributable to the unavailability of a more appropriate DSM diagnosis for youth with very severe, chronic irritability (see Spring and Carlson, volume I).¹ The most controversial change was the introduction of DMDD into the diagnostic manual. There are many valid criticisms of this action, with Evans and colleagues⁵⁵ providing a cogent and informative critique.

An important critique of DMDD is its overlap with ODD. Indeed, there are 3 diagnoses for which youth can qualify based primarily on their irritability or associated behavior: ODD, DMDD, and intermittent explosive disorder. Notably, the ODD diagnosis contains 2 dimensions, irritability and headstrong/defiant, that are dissociable in their longitudinal predictions and possibly in their genetic associations and treatment response (see Vidal-Ribas and Stringaris, volume I; Burke and colleagues, volume II).¹ The implications of these 2 commonly co-occurring dimensions for the ODD, DMDD, and IED diagnoses should be considered. A major consideration for the field concerns how these 3 diagnoses can be rationalized and integrated in the DSM.

A second major critique of DMDD is that thresholds for the frequency and severity of both outbursts and between-outburst mood were not derived empirically. Indeed, a necessary step in integrating ODD, IED, and DMDD is the acquisition of populationbased data to guide the development of empirical thresholds for pathologic irritability. With an appreciation of the developmental context in which these outbursts occur (see Crowell and also Dougherty and colleagues, Volume II),¹ important examples of such work are in a birth cohort of 9-year-old to 11-year-old youth,¹⁸ and in preschoolers.⁵⁶ Importantly, the latter work, along with other work²⁵ challenges limiting the DMDD diagnosis to children older than 6 years. Specifically, this work demonstrates than non-normative temper outbursts can be identified in preschoolers while illustrating the importance of developmentally sensitive criteria for irritability-based diagnoses.⁵⁷

Translational Opportunity: Frustrative Nonreward

The irritability construct provides important translational opportunities in its links to threat and frustration, 2 processes that can be modeled across species. Whereas our group’s clinical definition of irritability formed the basis for the preceding discussion, here we focus on our neuroscientific conceptualization of irritability, that is, aberrant approach responses to frustration or threat.² Specifically, this section focuses on using frustration tasks and fMRI to study individual differences in neural responses to frustration and thereby elucidate pathophysiological mechanisms of irritability (for a broader review, see Dickstein and colleagues, volume II).¹

Irritability and Frustrative Nonreward

The FNR paradigm was defined by Amsel in 1958.³ In recent adaptations,⁵⁸ rats are conditioned via instrumental learning to expect that, if they push a lever when a cue light is illuminated, they will receive a food reward. Then, the rat presses the lever when the light is on and yet no food is delivered, that is, the expected reward is withheld, goal attainment is blocked, and frustration is induced (Fig. 2A). Amsel³ noted that the normative response to frustration included increased motor activity and increased aggression. The FNR paradigm has been used in studies of rodents, nonhuman primates, and humans. Ongoing translational research bridges work on frustration in rodents and nonhuman primates to research on irritability in youth.

Fig. 2.

(A) A representation of the FNR paradigm first described by Amsel in 1958³ and adapted by Burokas and colleagues in 2012.⁵⁸ Rodents are conditioned to expect food reward if the lever is pressed when the cue light is on. Frustration is then induced by with-holding reward even if the lever is pressed at the correct time (blocked goal attainment). The emotional state of frustration is associated with increased motor activity and aggression, relative to baseline. (B) Brotman and colleagues² proposed a translational model for clinically relevant irritability in youth that includes an adaptation of Amsel’s³ FNR paradigm.³ Whereas Amsel’s model³ describes a normative response to frustration, Brotman and colleagues² suggest deficits that may contribute to an irritable child’s exaggerated and maladaptive response to frustration. These deficits may include aberrant instrumental learning or reward processing (see text). These aberrancies, in turn, result in an exaggerated emotional response (ie, increased anger and frustration in response to blocked goal attainment) and maladaptive behavior in the form of pathologically increased motor activity and aggression (ie, a temper outburst).

Our proposed model hypothesizes that irritable youth have exaggerated emotional and behavioral responses to FNR (Fig. 2B).² That is, when they experience blocked goal attainment, irritable youth experience a more marked high-arousal, negative affective state (ie, frustration, anger) than do their nonirritable peers. Related to this exaggerated emotional response, irritable youth show more extreme increases in motor activity and aggression, thus displaying developmentally inappropriate temper outbursts. These exaggerated responses could be due to deficits in instrumental learning, either because irritable youth have difficulty learning contingencies (deficits in the process of instrumental learning at baseline and/or when frustrated) or because their maladaptive behavior is rewarded (there are deficits in the content of instrumental learning, in that having an emotional outburst results in receiving the desired reward). In addition, aberrant behavioral responses could be initiated by exaggerated neural responses in the form of the prediction error that encodes the failure to receive an expected reward, or the receipt of an unanticipated punishment (Fig. 3).⁵⁹

Fig. 3.

Exposure-targeted model of pediatric irritability. Visual representation of proposed proximal processes that surround a clinically significant temper outburst in pediatric irritability, typically occurring within a timescale of minutes. White boxes denote key constructs in bold text. Within each box, additional text briefly describes the construct. Arrows denote direct or moderating influences of these components on one another. Boxes encompassed by red background indicate the primary targets of exposure-based cognitive behavioral therapy for irritability. Boxes encompassed by blue background indicate targets of parent interventions within this CBT. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) (Reprinted with permission from Ref.⁵⁹)

Reprinted from Kircanski et al., 2019, BRAT⁷⁶

Importantly, the translational perspective of FNR identifies potential targets for novel interventions (see Fig. 3, an adaptation of our original model focused specifically on treatment targets). These targets include the deficits in instrumental learning noted previously, some of which are addressed by parent training.⁶⁰ In addition, we now also highlight potential deficits in cognitive control (eg, motor inhibition, attention shifting) that might contribute to the inability to modulate responses to FNR.⁶¹ These targets have been incorporated into a novel treatment, currently being tested, that combines exposure-based techniques to frustrating stimuli with parent training techniques.⁵⁹

Frustrative Nonreward and Functional MRI

Although the literature is relatively limited, FNR paradigms have been used in normative and non-normative adult and pediatric samples, in behavioral, electroencephalogram, and fMRI studies (for a review, see Leibenluft⁶¹). The specific results of these studies must be viewed as preliminary, given small samples, methodological differences, and the lack of replication attempts. The discussion that follows will focus on the largest clinical fMRI study of frustration, with a focus not only on the results, but also on methods and complexities in this line of research.

The largest study of FNR in youth included 195 subjects (ages 12.9 ± 2.4 years) with DMDD, anxiety disorders, ADHD, or no diagnosis; thus, there was a range of irritability in the sample.⁶² The study used an adaptation of the Posner attention paradigm modified to induce frustration.^62–64 During the Posner attentional task, children are told that they will receive money for each correct response and lose money for each response that is incorrect or too slow (instrumental learning).

In the first block, children are given accurate feedback. Because the task is easy, they typically achieve 90% accuracy and receive frequent reward. The next block is rigged: on 60% of correctly completed trials, the child receives feedback that their response is too slow, and they lose most of the money won in the first block. fMRI analysis compares neural activity when the child completes the attentional task on the “N11” trial following frustrating versus nonfrustrating feedback. This contrast examines the clinically important question of whether attentional control differs in the frustrated versus nonfrustrated state, and whether such changes are associated with irritability. The statistical analysis focused on associations between irritability and neural activity on these 2 contrasts, controlling for the effects of anxiety or ADHD.

We found relatively strong associations between irritability and neural activity during the N11 trial, that is, when performing the attentional task after frustrating versus non-frustrating feedback. We highlight 3 clinically relevant aspects of the findings. First, increased irritability was associated with increased frontal-striatal activation in the anterior cingulate cortex, dorsolateral prefrontal cortex, inferior frontal gyrus, and caudate. These regions are implicated in cognitive control functions including motor inhibition and attentional control, suggesting the potential importance of such processes in potential treatment interventions. Second, some findings were most marked in younger subjects, indicating developmental effects on the neural mechanisms of irritability. And third, the findings associated with irritability were present across diagnostic categories (DMDD, anxiety disorders, ADHD); importantly, analyses based on diagnostic categories did not yield significant findings. This suggests that the mechanisms mediating irritability in DMDD, anxiety disorders, and ADHD are similar, consistent with the data noted previously showing genetic and longitudinal associations among these diagnoses (in contrast, see the findings of Wiggins and colleagues⁶⁵ where associations with irritability differed in DMDD vs BD).

Each fMRI paradigm, including each frustration paradigm, has limitations, so we use complementary frustration tasks that differ. For example, in a second frustration task, the order of frustration versus nonfrustration blocks is randomized, and a third task is designed to test whether irritable youth have deficits in instrumental learning at baseline and/or after frustration. Finally, our model suggests that irritability can be associated with aberrant responses to threat or frustration. This begs the question of whether there is heterogeneity among irritable youth (or adults) in the form of threat versus frustration-sensitive subtypes. We are currently testing this question in a study integrating data from a sample of youth who have completed both threat and frustration fMRI paradigms.

SUMMARY

There are several take-away messages from this review of irritability, viewed particularly from a translational perspective. First, irritability is one of many constructs that can be used to study emotion dysregulation. It has reasonable external validity, with more data accumulating regarding its biology, genetics, and response to specific treatments. Further, it has good longitudinal stability and ability to predict clinical course, and these validators are being examined with increasingly sophisticated statistical techniques. Importantly, to the extent that irritability is conceptualized as aberrant responses to frustration and/or threat, it is amenable to cross-species research, that is, responses to both frustration and threat can be assessed in humans and animals using parallel tasks.

Research suggests that phasic irritability (temper outbursts) may be a distinct construct from tonic irritability (grouchy mood between outbursts). Important questions include whether these constructs have differential longitudinal outcomes or response to treatment, and the degree to which the severity of the tantrums in terms of behavior, duration, or pervasiveness impacts on outcome or treatment response. EMA may help answer these questions, given the difficulty of differentiating tonic versus phasic irritability by retrospective report. Such research could complement data from large epidemiologic studies, and from samples enriched for irritability. Epidemiologic studies are needed to determine thresholds for clinically significant irritability and enable the rationalization and integration of irritability-related diagnoses in the DSM.

The hypothesis that irritability reflects exaggerated negative responses to frustration suggests that standardized tasks that induce frustration can be used to study the neural mechanisms mediating irritability in youth. Such studies suggest associations between increased irritability and increased fronto-striatal activity when children attempt to complete an attentional task immediately after frustration. This work informs the development of a novel psychotherapeutic approach to irritability, currently being tested. The therapeutic approach includes the child experiencing graded exposure to frustration and learning to increase his or her cognitive control capabilities in order to be able to inhibit temper outbursts. Concurrently, parents receive training teaching them to reward the child’s positive, but not negative, behavior.

Chronic irritability in youth is associated with severe impairment and adverse outcomes, and there is also considerable burden on their families. Hopefully, careful and systematic conceptualization and research focused on their emotional and behavioral difficulties will ultimately help to alleviate the considerable suffering of these children and their families.

KEY POINTS.

• Irritability is a subtype of emotion dysregulation that is characterized by increased proneness to anger, relative to peers, and is a suitable focus for translational research.

• Chronic, severe irritability is not a pediatric presentation of bipolar disorder. Instead, it is longitudinally associated with unipolar depression and anxiety and genetically associated with unipolar depression, anxiety, and attention-deficit hyperactivity disorder.

• Frustrative nonreward is a construct that can be evaluated across species to understand mechanisms underlying irritability.

• Research on brain mechanisms mediating aberrant responses to frustration in youth can guide the development of novel interventions.