Does circadian dysrhythmia drive the switch into high‐ or low‐activation states in bipolar I disorder?

Abstract Objectives Emerging evidence suggests a role of circadian dysrhythmia in the switch between “activation” states (i.e., objective motor activity and subjective energy) in bipolar I disorder. Methods We examined the evidence with respect to four relevant questions: (1) Are natural or environmental exposures that can disrupt circadian rhythms also related to the switch into high‐/low‐activation states? (2) Are circadian dysrhythmias (e.g., altered rest/activity rhythms) associated with the switch into activation states in bipolar disorder? (3) Do interventions that affect the circadian system also affect activation states? (4) Are associations between circadian dysrhythmias and activation states influenced by other “third” factors? Results Factors that naturally or experimentally alter circadian rhythms (e.g., light exposure) have been shown to relate to activation states; however future studies need to measure circadian rhythms contemporaneously with these natural/experimental factors. Actigraphic measures of circadian dysrhythmias are associated prospectively with the switch into high‐ or low‐activation states, and more studies are needed to establish the most relevant prognostic actigraphy metrics in bipolar disorder. Interventions that can affect the circadian system (e.g., light therapy, lithium) can also reduce the switch into high‐/low‐activation states. Whether circadian rhythms mediate these clinical effects is an unknown but valuable question. The influence of age, sex, and other confounders on these associations needs to be better characterised. Conclusion Based on the reviewed evidence, our view is that circadian dysrhythmia is a plausible driver of transitions into high‐ and low‐activation states and deserves prioritisation in research in bipolar disorders.

(1) delineation of "vascular depression" among older adults with a first onset of depression in later life after specific cerebrovascular lesions 1 ; (2) onset of depressive disorders after specific infections 2 or the introduction of immune-active therapies 3 ; and (3) recognition of "atypical" mood disorders in young adults with antineuronal antibodies. 4 The latter discovery has led to the use of immune therapies targeted to underlying pathophysiology. 5 One pathophysiological mechanism that may be relevant to major mood disorders, and particularly bipolar I disorder (BD-I), is disturbance in the 24-h circadian system. [6][7][8] In recent years, the field of "Circadian Medicine" has made major contributions to our scientific understanding and treatment of many illnesses, including dementia, cancer, and cardiovascular disease, 9 with the Nobel Prize in Physiology or Medicine awarded for the delineation of the molecular mechanisms controlling the 24-h circadian clock. While the relevance of perturbed circadian rhythms (or "dysrhythmias") to mood disorders has been long recognised, 10 more recent neurobiological, interventional, and longitudinal studies justify a more considered re-evaluation of the potentially causative nature of these associations, specifically regarding whether circadian dysrhythmias drive the switch into "activation" states in BD-I.
By "activation" states, we refer to the linked phenomena of objective motor activity and subjective energy, and critically, to intraindividual change in these phenomena. 11 Conceptually, we first deconstruct the category of BD-I into two broad constituent states and a transition process: (1) a "high motor activation" (manic) phase; (2) a "low motor activation" (depressed) phase; and (3) a tendency to switch between these abnormal "activation" states. In this framework, the primary focus is on activation, while subjective mood is considered separately. 11 This phenomenological deconstruction is supported by factor analytic, 11 family, [12][13][14] and clinical studies, which indicate that activation and mood may vary independently. Such decoupling of activation and mood is exemplified in mixed states (e.g., high activation and low mood in "dysphoric mania"; high activation and low mood in "agitated depression"); although these are not the focus here. 15 Our focus on activation as a core feature of BD-I is consistent with DSM-5's recognition of changes in activity and energy (alongside mood) as a criterion A symptom, and recent publications about the potential primacy of activation in BD, which have highlighted the need for investigation of the physiological substrates of the transition between activation states, to which circadian rhythms may be relevant. 11,[16][17][18][19][20] "Circadian dysrhythmias" (variously defined) are commonly reported in individuals with BD-I and are associated with a range of core clinical features. 21 We use circadian dysrhythmia as an umbrella construct that represents a "change in one or more aspects of a circadian cycle's morphology". 22 We recognise that multiple phenotypes could be relevant to activation states in BD (e.g., phase delay, phase advance, high fragmentation, blunted amplitude, internal desynchrony, arrhythmia), and we note that most studies in this area have examined actigraphy-based estimates of phase, amplitude, and fragmentation. In this article, we consider whether circadian dysrhythmias may be a causative driver of the switch into high and low motor activation states in BD-I, rather than mere correlates or epiphenomena.
We undertook a selective/narrative review of the evidence of the association between circadian dysrhythmias and changes in activation states in BD-I. We summarise the evidence in tables that examine four key questions, and finally discuss the available evidence together. Our target questions were as follows: 1. Are natural or experimental exposures that are associated with circadian dysrhythmia also associated with the switch into highor low-activation states? (Table 1) 2. Are circadian dysrhythmias associated with the switch into highor low-activation states? (Table 2) 3. Do interventions that affect the circadian system also affect highor low-activation states? (Table 3) 4. Are associations between circadian dysrhythmias and high-or low-activation states confounded by "third" factors? ( Table 4) Our selection of evidence for this review was based in part on our clinical experience and associated priors (e.g., potential links between infection, circadian disturbance, and BD), literature searches using key terms (e.g., "circadian rhythms", "actigraphy", "motor activity", "bipolar", "mania", "depression", "clinical trial"), and searches through our personal files. We aimed to primarily include studies that explicitly report on motor activation; however, we also include studies that examined biological circadian rhythms (e.g., melatonin), and studies of patients that may not have been included based on motor activation. We note that the scope of this review is focused, and we point interested readers to a recent comprehensive review 8 of additional areas that fell outside our scope (e.g., preclinical cellular and animal models, molecular genetic studies, studies not focused on motor activation).

| DISCUSS ION
In this narrative review, we have outlined evidence relevant to our hypothesis that circadian dysrhythmias have the capacity to drive switches into high-or low-activation states in BD-I. While we postulate that circadian dysrhythmias can causally increase the likelihood of these transitions, it is highly likely that this risk is conditional on other risk factors, such as genetic risk for BD or biological and environmental factors (e.g., sensitivity to light, local variation in light or other seasonally patterned factors). In other words, circadian dysrhythmias may be a component cause of variation in the course of activation states in BD-I. 23 It is also important to note that while we focused our efforts on bipolar disorders (and BD-I specifically), there is evidence that circadian dysrhythmias may play a role in the course of other mental disorders (e.g., psychotic disorders, depressive disorders), 24 (Table 1) and in treatment studies (Table 3). Very few studies used direct measures of circadian timing, and conclusions about central timing based on real-world actigraphy data are hampered by their limited concordance with central measures (e.g., dimlight melatonin onset) and controlled settings (e.g., constant routine), and the susceptibility of actigraphy to various masking effects (e.g., voluntary behaviour, social activity, work schedules). Finally, we acknowledge that our review covers a focused area (human studies relevant to motor activation in BD-I) in the wider context of chronobiology and mental disorders, for which there are reviews 8,26 on preclinical models, 27-29 neurobiology, 7,30 phenomenology, 31 measurement, 17,31 pharmacology, 28 modelling, 32 and genetics. 33,34 TA B L E 1 Associations between transitions into high-or low-activation states and natural or experimental exposures capable of disrupting the circadian system. • Links between infection and activation states may be due to non-circadian effects (e.g., immune activation) 76 • Lack of objective measurement of circadian dysrhythmias TA B L E 2 Associations between estimated circadian dysrhythmias and the course of low-and high-activation states.

Switch into high activation ("Mania") Switch into low activation ("Depression") Comments and potential confounds
Rest-activity rhythms • Intra-daily variability of rest/ activity rhythm associated with relapse (manic, hypomanic, or mixed episode) 52 • Null finding of a within-person association between rest/activity rhythm and days spent in a manic/ hypomanic episode 53

TA B L E 3
Associations between interventions that act on the circadian system and high-and low-activation states in BD-I. and low (depression) activation states and should provoke study of these exposures using prospective, hypothesis-testing designs.

Switch into high activation ("Mania") Switch into low activation ("Depression") Comments and potential confounds
The next most convincing factor is that interventions that target the circadian system (e.g., bright light therapy, dark therapy, melatonergic agents, lithium) can lead to resolution of mania or depression. While lithium has strong evidence for amelioration of mania, and reduction in relapse, its exact effects on circadian systems remain to be fully characterised. However, there is substantial TA B L E 4 Are associations between circadian dysrhythmia and high-and low-activation states explained by a third common factor?

Age and sex
• Several studies show associations between BD-I and circadian dysrhythmia independent of age and sex 18 • One study shows an association between transition into mania and objective circadian dysrhythmia, adjusted for age/sex 52

Age and sex
• Several studies show associations between BD-I and circadian dysrhythmia independent of age and sex 18 • One study shows an association between transition into depression and objective circadian dysrhythmia, adjusted for age/ sex 52 • More studies needed that account for age and sex

Neurodevelopmental impairment
• Circadian rhythm sleep/wake disturbance (e.g., delayed sleep phase) are observed in samples excluding participants with low IQ or history of head injury 97 • This evidence comes from a study of a mixed diagnostic sample 97 • Few studies appear to examine these types of impairments

Genetic factors
• Mixed evidence regarding associations of circadian genes and BD (broadly defined) 33,98 • Some evidence for genetic associations among BD-I pedigrees and circadian dysrhythmia (blunted rest/activity rhythm) 99 • No association between polygenic risk score (PRS) for low amplitude of rest/activity rhythm and BD, but a significant association with mood instability 58

Genetic factors
• Mixed evidence regarding associations of circadian genes and BD (broadly defined) 33,98 • Some evidence for an association among a seasonal pattern of depressive episodes and circadian genes 100 • Lack of evidence regarding associations of manic and depressed states and circadian dysrhythmias, accounting for genetic factors evidence of the circadian effects of lithium across a range of human and animal studies (e.g., cultured cells, healthy humans, animal models, patient-derived fibroblasts). 27,39,40 For depression, there is also evidence that melatonergic agents (e.g., agomelatine) and bright light therapy may have significant benefits [41][42][43][44][45] ; however, this evidence is quite mixed, [44][45][46][47][48]  and circadian phenotypes. [54][55][56][57] For example, a study using a polygenic risk score (PRS) for actigraphy-based relative amplitude found no association with BD, but a significant association with mood instability. 58 A genome-wide association study (GWAS) of >40,000 cases with BD reported positive genetic correlations between BD and sleep phenotypes (insomnia, daytime sleepiness, sleep duration, daytime napping, "getting up in morning") but not chronotype. 56 Moreover, there were bidirectional (putatively causal) associations between BD and longer sleep duration, and a unidirectional (putatively causal) association between BD and a lower likelihood of being a morning person. 56 However, some GWASs have reported mixed findings, with significant genetic correlations between BD and some sleep or circadian phenotypes (e.g., sleep duration) but not others (e.g., insomnia, chronotype). 55 A casecontrol study observed differential associations among BD subtypes and polygenic liability to sleep duration and insomnia, but not for chronotype. 57 While mounting evidence shows shared genetic risk for BD and sleep phenotypes, we note that the role of genetic and biological factors (e.g., neurodevelopment) linking circadian phenotypes and BD is less clear, and should be explored further in genetic, twin, family, and prospective studies, particularly those following individuals through developmental phases of peak risk of BD.
Considering the evidence reviewed, our view is that circadian dysrhythmia is a plausible driver of transition into both high-and low-

CO N FLI C T O F I NTE R E S T S TATE M E NT
IBH is the Co-Director, Health and Policy at the Brain and Mind