Medical Hypotheses 82 (2014) 581–588

Contents lists available at ScienceDirect

Medical Hypotheses journal homepage: www.elsevier.com/locate/mehy

Delayed mood transitions in major depressive disorder Jakob Korf ⇑ University of Groningen, Centre of Psychiatry, Groningen, The Netherlands

a r t i c l e

i n f o

Article history: Received 12 September 2013 Accepted 16 February 2014

a b s t r a c t The hypothesis defended here is that the process of mood-normalizing transitions fails in a significant proportion of patients suffering from major depressive disorder. Such a failure is largely unrelated to the psychological content. Evidence for the hypothesis is provided by the highly variable and unpredictable time-courses of the depressive episodes. The main supporting observations are: (1) mood transitions within minutes or days have been reported during deep brain stimulation, naps after sleep deprivation and bipolar mood disorders; (2) sleep deprivation, electroconvulsive treatment and experimental drugs (e.g., ketamine) may facilitate mood transitions in major depressive disorder within hours or a few days; (3) epidemiological and clinical studies show that the time-to-recovery from major depressive disorder can be described with decay models implying very short depressive episodes; (4) lack of relationship between the length of depression and recovery episodes in recurrent depression; (5) mood fluctuations predict later therapeutic success in major depressive disorder. We discuss some recent models aimed to describe random mood transitions. The observations together suggest that the mood transitions have a wide variety of apparently unrelated causes. We suggest that the mechanism of mood transition is compromised in major depressive disorder, which has to be recognized in diagnostic systems. Ó 2014 Elsevier Ltd. All rights reserved.

Introduction The life-time prevalence of major depressive disorder (MDD) in the Western population is 10–20%. In approximately 50% of the cases MDD is diagnosed and treated by the general practitioner, while severe depressions with or without psychiatric co-morbidity are referred to psychiatrists. If not adequately treated, depression may become a life-threatening psychiatric condition. Indeed, suicide rate is high in depression. Antidepressant drugs, including serotonin reuptake inhibitors, are often the treatment of first choice, but their effectiveness is – at least in large cohorts – often little more than a placebo treatment [1–3]. In addition, the search for specific diagnostic markers for depression to enable better therapeutic results is rather unsuccessful [2,3]. Concerning psychotherapies there is little if any evidence of the best therapeutic efficacy among seven options including cognitive behavior therapy, interpersonal psychotherapy, behavioral activation, problem solving therapy, psychodynamic therapy, nondirective counseling, or Abbreviations: DSM, diagnostic and statistical manual of mental disorders; ECT, electroconvulsive therapy; MDD, major depressive disorder; SSRIs, selective serotonin reuptake inhibitors. ⇑ Address: UCP Rm 4.14 CG 10, UMC Groningen, P.O. Box 30.001, 9700 RB Groningen, The Netherlands. Tel.: +31 592 541974. E-mail addresses: [email protected], [email protected] http://dx.doi.org/10.1016/j.mehy.2014.02.015 0306-9877/Ó 2014 Elsevier Ltd. All rights reserved.

social skills training [4 and references therein]. And, finally, authors have pointed to the scientific weaknesses of current classification systems, including the Diagnostic and Statistical Manual of Mental Disorders [DSM; 1,3]. Taken together, these notions illustrate the limited progress made of a scientific and clinical concept of MDD over the last six decades [1]. This lack of progress has often been attributed to the complexity of the brain and to the difficulties faced when trying to extract relevant information from the living brain. Another major problem is the presumed causal relationship between the nature of psychopathology and successful treatment. Does, for instance, the psychological content of depression, e.g., the depressive feelings, determine the timing of recovery, or do some underlying mechanisms govern the time course of depression? The therapeutic implications might than differ: the first mechanisms appeal primarily to psychotherapeutic approaches whereas the latter suggest interventions associated to some underlying, presumably neurobiological, mechanism or pathology. The latter idea implies that therapeutic efficacy is primarily limited by the probability of mood-normalizing transitions. The hypothesis forwarded here elaborates on this idea and in particular on the time structure of MDD and associated depressive episodes. In the present report I use the Diagnostic and Statistical Manual of Mental Disorders version DSM IV-R and not the most recent DSM

582

J. Korf / Medical Hypotheses 82 (2014) 581–588

V, because far most of the reviewed studies are based on earlier versions [3,5]. Generally speaking MDD is considered the collective impact of external challenges, cultural and social environment, personal experiences and biological disposition and is an amalgam of symptoms. A tacit assumption of the DSM is that MDD is an underlying pathology that manifests itself in a variety of symptoms, that might be personal with loose inter-symptom connections. The duration of the depressive condition is an essential characteristic: at least 2 weeks according to the DSM classification. Shorter periods of depression are considered as non-pathological and do not deserve psychiatric attention. We discuss some implications of our hypothesis for the scientific character of diagnostic systems. Hypothesis The hypothesis defended is that the core pathology in a significant proportion of patients suffering from MDD is the failure of mood-normalizing transitions which is not necessarily related to a presumed psychological content, i.e., sad thoughts. I describe and discuss several examples on the timing of mood transition, which are summarized in Fig. 1. Fast mood transitions may occur following joyful or sad experiences in everyday life. Clinical and case reports of fast depressiogenic transitions have been observed following such interventions as deep brain stimulation, sleep deprivation and subsequent short naps. Other arguments are provided by clinical and experimental psychopharmacological interventions and epidemiological data surveys (depicted in Figs. 2 and 3). The transition-hypothesis’s potential is therefore further underlined by analytical models (summarized in Fig. 4). Fast mood switches in non-MDD subjects A core assumption of our hypothesis is that transitions of the depressive mood in MDD might be fast, as is occasionally observed

during deep deep-brain stimulation and in ultra-rapid mood-cycling patients. In normal life joyful or sad experiences may result in fast mood transitions within minutes or even seconds. Mood may become depressed after a message of the death of a beloved, or conversely, elated by a successful effort in one’s career or in sports. These transitions may normalize in minutes or days, depending on the experienced impact. In clinical practice depression is often co-morbid with somatic conditions such as cancer, cardiovascular disease or diabetes. In these cases depression is commonly not diagnosed as MDD, but rather as depression associated to a general medical condition [5]. The associated depressive feelings may readily dissolve following successful treatment of the somatic syndrome. Fast mood transitions have been observed in non-MDD patients. For instance, electrical stimulation of the subthalamic nucleus, aimed to alleviate tremor in Parkinson’s disease, evokes a depressed mood within 5 s, which than disappeared within 30 s after cessation of the stimulation [6–9]. In some patients the depressive mood was induced repeatedly and was accompanied by suicidal ideation or attempt. Electrical stimulation does not imply that the neuronal pathway becomes more active: rather it impairs the integration of the stimulated pathway in a functional neuronal network. Hence, electrical stimulation might be seen as a reversible block of neuronal pathways, here perhaps dopamine neurons. Together these case reports demonstrate that a pathological or near-pathological depressive state of mood might precipitate and dissolve within minutes. The question remains open whether the very fast mood transitions during deep brain stimulation are characteristic for some subjects or whether they might be evoked in (almost) every individual. Bipolar patients with manic or depressed episodes as short as a single day or a few days, were described more than 25 years ago [10]. About 25 cases of ultra-rapid cycling and 15 cases of ultradian cycling have been reported. Frequent short periods of depression (1–4 days) were noted in a systematic study of 203 bipolar patients

Fig. 1. Time frames of mood transitions. All time frames considered in the text are shown. Within-minute transitions include normal life experiences, deep brain stimulation and naps following sleep deprivation (SD). Antidepressive responses following various experimental drugs (ketamine, scopolamine, cortisol/hydrocortisone), the first two electroconvulsive treatment (ECT) sessions and the depressiogenic response to acute tryptophan depletion cause lasting or transient mood effects within hours or a few days. In the general population the time-to-recovery course is described with an exponential or a Weibull decay model: in these cases the is no average length. According to the Weibull model, placebo and antidepressant drugs may partially alleviate mood in some patients to about 50% (of depression scores) in one or two weeks, whereas the majority of patients follow a gradual antidepressant trajectory. Right panel signifies the fraction of depressive patients unresponsive to interventions. Details in the main text.

J. Korf / Medical Hypotheses 82 (2014) 581–588

583

Fig. 2. Time course of depressive episodes. (A) NEMESIS cohort, general Dutch population 250 subjects [32,33]; (B) INSTEL cohort of recurrent depression 270 subjects [34]; (C) cohort of almost 3000 subjects from the general population (the Canadian Community Health Survey Mental Health and Well Being population [37]. Solid marks indicate observations. Graphs are fitted lines of mono-exponential decay functions (estimated by an overlay technique and visual inspection, based on quantitative modeling of the NEMESIS and INSTEL studies). The Canadian study was originally fitted with a Weibull model, which describes the entire time course better than the exponential decay function. Nevertheless, the fit quality of both models in the Canadian study was rather similar during the first 100 weeks. All three cohorts show rapid recovery during the initial periods, though the decay times differed. Notice the differences in length of the episodes.

Fig. 3. Time patterns of improvements. Time courses of Hamilton depression scores of placebo treated subjects, and of responders and non-responders receiving active drug treatment [data redrawn from 48]. The time-points are modeled with exponential decay models. The models of the responders and placebo cohorts were similar to that of the time-to-recovery from depression in the general population (Fig. 2).

using daily self-reported mood ratings collected during an average of ½ year of observation [11]. Overall severe symptoms were noted in about 25% during one-day depressive episodes. Ultra-rapid cyclers are not a DSM-diagnostic category and might therefore be under-diagnosed. Fast transient sub-syndrome symptoms and mood-instability parameters might predict functional recovery both in bipolar disorders [12,13] and MDD. These observations together illustrate that depression-inducing or -relieving transitions might be fast (within minutes or days), irrespective of the psychological content of the feelings and do not necessarily originate in psychodynamic mechanisms. Sleep deprivation and naps Manipulations of the sleep-wake cycle, whether in duration (total or partial sleep deprivation) or timing (partial sleep

deprivation, phase advance), have profound and rapid effects on depressed mood in about 50% of all diagnostic subgroups of affective disorders (more than 1700 patients reviewed) [14,15]. The effects of sleep deprivation are rather unstable: in about 50% of the patients the antidepressant effects have disappeared after one night of sleep. Short naps or even short bouts of microsleep the day after sleep deprivation precipitate both positive and negative mood responses, with morning naps being more detrimental than afternoon naps [16]. Suppression of microsleep by flumazenil did not sustain the antidepressant effect of sleep deprivation [17], thus questioning the role of sleep. The response to one night of sleep deprivation does not predict the response to a next deprivation [14,18]. Relapses appear to be less when combined with concomitant treatment such as, lithium salts (extending the effects over four days), serotonin selective antidepressants, bright light, or a subsequent phase advance procedure [14]. Diurnal and day-to-day mood variability predict both short-term response to sleep deprivation and long-term response to antidepressant drug treatment [14,19], suggesting that the propensity of mood transitions facilitates the therapeutic efficacy of the intervention. The antidepressant response of sleep deprivation has been compared to that of the glutamatergic antagonist ketamine ([20], to be discussed later). In conclusion: most of the well-designed sleep studies emphasize, is that in (at least some) MDD patients the depressive mood may switch within hours and -sometimes-even within a few minutes. Direct brain stimulation The general clinical impression is that ‘‘in some situations the action of electroconvulsive treatment (ECT) may be more rapid than that of psychotropic drugs.’’ Indeed, a 50% improvement on the depression scores (Hamilton) after one or a few sessions of ECT has been reported [21–25]. An illustrative example: 47 inpatients with MDD were randomly assigned to twice weekly bilateral ECTs, brief pulse plus one simulated treatment per week or to a schedule of administration three times per week. Already after two weeks, a reduction of the scores by approximately 30% in the three times weekly schedule was noted. Thus predicting later success of the treatment regimen [25]. Among the responders, about 25% of the overall improvement in the depression scores

584

J. Korf / Medical Hypotheses 82 (2014) 581–588

Fig. 4. Basic features of 3 transition models. (A) Dynamic system critical model (DSCT). The relative change over time of the significance of two competing brain states or processes: the depressed and recovered states, respectively. This depressed state gradually loses influence, reaching a critical time point followed by a relatively fast transition to the recovered state. The intermediate state is characterized by fluctuation (instabilities), pointing to an increased propensity for a mood transition. (B) Network connectivity model (NCT). The left panel is an example of a symptom network showing a predominant position of depressive symptoms. The right figure shows increasing efficacy of the interaction of the components of the network due to an increasing connectivity in the network. An increasing connectivity might ultimately lead to an increasing instability of the network, followed by a collapse. This way, the depressive state of the brain might dissolve to be replaced by one or more non-depressed states. (C) Quantum mind/brain model (QMBT). The idea is borrowed from quantum–mechanical theories describing the transition of a wave function into a particle, meaning that a non-localized state (the wave) may suddenly (and randomly) collapse to become a localized particle. In the present terminology, depression is considered an undefined state of the brain that might be dissolved by applying challenges or interactions, which could be any routine or experimental treatment mentioned in this essay. Figure A (redrawn from [56]; part of figure B (from [58]).

was contributed by the first real ECT. Early response to ECT predicted success at the end of the treatment: a decrease of the Hamilton scores by about 30% after two ECTs was significantly lacking in the late- and non-responders. These studies underscore the idea that the anti-depressive effects of ECT might become significant after a few convulsions.

During the last decade the efficacy of trans-cranial magnetic seizure therapy, deep brain and vagus stimulation have been explored, but their clinical efficacy has still to be established [22,23,26]. Their beneficial responses appear to be slow. Both brain stimulation techniques have to be optimized by, for instance, identifying effective brain locations and stimulus parameters [26,27].

J. Korf / Medical Hypotheses 82 (2014) 581–588

This section shows anti-depressive responses to ECT already after a few treatments which might predict later therapeutic success. Experimental drugs Several studies have claimed fast anti-depressive responses to experimental drug treatment paradigms. Five examples are discussed here. Acute antidepressant effects of intravenous hydrocortisone (cortisol) and bovine corticotropin releasing hormone or saline infusions were tested double blind in 22 patients who met criteria for non-psychotic MDD [28]. Only acute hydrocortisone infusion was associated with a rapid and robust reduction in depressive symptoms (37% of the 21-item Hamilton depression scores). Intravenous infusion of a sub-anesthetic dose of ketamine (a glutamate NMDA antagonist) causes a significant but temporary relief of symptoms in otherwise treatment resistant patients with MDD [20,29]. After a single dose of ketamine to more than 80% of the patients responded with a mean reduction to halve of depression scores (Montgomery Asberg) already after 2 h, with a further reduction to 20% after 24 h. Compared to placebo treatment, the effect of ketamine was already significant within 110 min and continued to remain significant for one week [20,29]. Little or no response to the first dose of ketamine may predict poor response to subsequent dosing [29]. Ketamine and some other explored rapidly acting antidepressant drugs are directly or indirectly acting N-methyl-D-aspartate receptor antagonists, thus impairing cerebral glutamate neurotransmission. Duncan and Zarate [20] propose that ketamine could act as antidepressant by increasing slow wave EEG-activity similar to that of sleep deprivation. Intravenous scopolamine (an antimuscarinic drug) was tested in 48 depressive outpatients in two double-blind crossover designs [30]. Anticholinergic side effects such as a dry mouth, blurred vision, lightheadedness and dizziness were noticed in almost all subjects. The depression scores (Montgomery–Asberg) decreased following scopolamine and significantly less after placebo. The group receiving scopolamine first showed a 32% reduction, compared to a change of 6.5% under placebo. Individuals reported improvements already the evening or the morning after the scopolamine administration. The cohort analysis revealed highly significant antidepressant effects within three days and lasting 10 days. Although highly significant, the overall antidepressive response was about 50% on the depression score, yet insufficient for a clinical recovery. Mood transitions can be induced by lowering blood tryptophan levels, thereby depleting brain serotonin [31]. Depressive symptoms (Hamilton scores) develop transiently in patients with a history of depression or in responders on SSRIs within hours and dissolve rapidly after normalization of the blood tryptophan levels. These results illustrate that low brain tryptophan precipitates depressive symptoms only in vulnerable subjects. The here reviewed reports show that some experimental drugs may cause rapid (often within hours) anti-depressive responses in MDD. In this respect, ketamine is most promising. The various mood-normalizing interventions described in the present and preceding 3 sections lack clear-cut common underlying mechanism of action. Time-to-recovery from depression The following 3 sections summarize the time course of depression in the general population and during antidepressant interventions. The present section covers epidemiological studies from the

585

Netherlands, Finland and Canada on the duration of MDD episodes in the general population. The Dutch study (The Netherlands Mental Health Survey and Incidence Study [32]) consisted of 250 subjects suffering from an episode of MDD (from a population of about 7000), who were followed for 24 months [32,33]. The number of depressed subjects versus the duration of the episode was modeled and a mono-exponential decay model gave the best fit with a residual fraction of 20% remaining depressed (Fig. 2a). The model fitted irrespective of psychiatric (anxiety, dysthymia) or somatic co-morbidity, gender or severity of the depression. The model suggests, that women develop a depression of shorter duration and that severe depression shows longer decay times. The exponential model predicts very short (days rather than weeks) episodes of depression, implying the arbitrariness of the DSM inclusion criterion of two weeks. The single exponential model fitted both first and subsequent episodes of recurrent MDD [34] (Fig. 2b). There was no relationship between the duration of subsequent depressive episodes in the same patient [34]. The exponential time-to-recovery model was consistent with random mood fluctuations [32], i.e., without a clear and general cause. In the Finnish study, exponential-like time courses of depression in the primary care (89 subjects) were observed with a fraction of about 30% remaining depressed after 18 months [35]. Inpatients exhibited an exponential-like recovery as well [36]. Neither of the latter data was modeled. The Canadian Community Health Survey of Mental Health and Well Being included nearly 3000 depressed subjects who met the DSM criteria for MDD. The 12 month prevalence of depression was assessed retrospectively [37]. Here an exponential model was compared with a Weibull function (Fig. 2c). Both models show that during the first six months the recovery from depression in a large majority of subjects progresses at a high and (nearly) constant probability whereas they deviate at later time intervals [37]. Another example of fast mood transitions is covered by the diagnosis recurrent brief depression referring to depressive episodes occurring at least once a month and lasting for only a few days [38]. This condition was recognized in more than 30% of the patients in the general practice [39]. This diagnosis illustrates again that depression may precipitate and resolve in a few days. These epidemiological studies converge in the idea that the probability to recover from MDD is highly variable. The models used imply that the duration of the depressive or intermittent episodes does not affect the probability to recover neither in a cohort, nor in the same subject. Both epidemiological observations and the well fitting exponential or near-exponential models imply that the majority of depressive episodes in a general population is short. These various observations and our analyses are in line with the concept of random mood transitions.

Momentary mood fluctuations Most psychiatric research on MDD relies on mood indices measured only a few times per week or so. If mood does indeed fluctuate rapidly moment-to-moment measurements might reveal supporting observations. The potential of this approach has been recognized in single-subject studies and, more recently, in cohorts as experience sampling or ecological momentary assessment. First the single subject approach. The time structure of depression was modeled in a longitudinal study in two women: one control and another with a uni-polar recurrent depression. Sadness was assessed hourly (10 times per day) by self-monitoring during six months [40]; these data were then modeled. A strong periodic component which existed only within the depressed subject was delineated, pointing to a low-dimensional chaotic process. In the non-depressed subject no such periodicity was identified.

586

J. Korf / Medical Hypotheses 82 (2014) 581–588

Frequent assessments in real time applied to cohorts of MDD (recent reviews [41–43]) allow both inter- and intra-person comparisons and the impact of biological and environmental variables. The immediate impact of external events (often daily hassles) on positive (interested, excited or alert) or negative (distressed, irritable, nervous) internal states were monitored. On average MDD patients reported less enjoyment of everyday activities, more complaints, and a poorer quality of life, than healthy controls [41]. Hence the negative response to daily hassles seems characteristic for MDD. Interestingly, such negative reaction appears to be determined both inherited vulnerability and to adverse (childhood) experiences [42]. The frequent sampling approach has also been explored to predict therapy response: positive affect in the first week of antidepressant treatment signifies later therapeutic response [43,44]. It appeared that indicators of slowing down of reactivity to environmental and other challenges are also predictive of future transitions in depression. Persons who are more likely to have a future transition to the depressive state, mood dynamics are slower [44]. Apparently both the speed and the magnitude of affective reactivity predict both susceptibility and clinical improvement. Depressed patients are more likely to respond to their treatment if they have greater positive affect persistence at baseline [43–45]. An underlying assumption of the moment-to-moment approach is that apparently minor events or experiences have an immediate effect on mood regulation. Positive and negative affect responses appear to be confined to different dimensions, suggesting unrelated underlying brain mechanisms [[45] and references]. The application of moment-to-moment scoring allows to developing a construct how the interactions of personal factors might cumulate up to depression [‘building blocks’ [42]], and which subjective factors have to be influenced to relieve his/her depression. Both cohort and individual studies have been subjected to mathematical network modeling (see section on modeling): labile connections might cause MDD. Together the symptom interplay approach may result in a syndrome that fulfills the (broad) criteria of MDD. The depressive state might dissolve rapidly when a critical number of the presumed subjective building blocks [42] are eliminated.

(time-to-recovery from depression). The time course of depression scores during treatment was monitored in a naturalistic study with 1014 inpatients (no placebo) [50]. After the first two weeks about half of the total reduction of the depression scores had already been reached. In another recent large study [51], the trajectories of the Montgomery–Asberg scores were analyzed (latent class) in individual patients. The individual scores showed high variations from week to week, often switching between near-recovery to severe depression and vice versa. Despite these variations, the individual scores could be divided into two groups. One (about 75%) showed a gradual decrease of the scores with a halve life of six weeks, whereas the remaining group (25%) exhibited a biphasic decline with an fast decline of the scores in the first 2 two weeks (to about half of the initial scores). The latter time course was rather similar to the exponential declines observed in the epidemiological (Fig. 2) and the time-to-recovery studies (Fig. 3). The time course of the net-drug versus the placebo response in MDD was modeled to test whether the final effect of the antidepressants could be predicted from the Hamilton-17 items scores at two or four weeks [52,53]. A Weibull/linear model accurately described the population and individual time course of the depression scores and the placebo response [52]. This model predicts later placebo and drug responses already at the onset of the intervention. The individual depression scores shown in the Gomeni studies [52,53] point to a fast response to antidepressant drugs that matches the data of Figs. 2 and 3. Together most of the recent meta-analysis of placebo-controlled, randomized trials reveals that by the end of the first week the response to antidepressants was significantly larger than to placebo. The early decline of depression scores can be modeled with exponential or near-exponential functions. Such exponential recovery-curves imply that a stochastic (or random mood) model applies to placebo and a variety of interventions, irrespective of treatment. Mood fluctuations observed in the first week of treatment predict often later therapeutic success. A remarkably similar conclusion was reached in studies on sleep deprivation, electroconvulsive treatment and with the momentary assessment approach.

Modeling fast mood transitions Antidepressant response rate Antidepressant medications are generally considered to have a delayed onset of action, but this conventional wisdom has been challenged during the last decade. This section describes examples showing significant responses within a few days or weeks after the initiation of an antidepressant intervention. This summary on the differentiation of the placebo antidepressant drug effects is based on meta-analyses and exemplary studies [e.g., [46–49]]. One of the few examples where antidepressant medication and cognitive behavioral therapy was compared is that of Tadic´ and coworkers [46] with data from 223 patients of a 10-weeks randomized, placebo-controlled trial. In both the sertraline and cognitivebehavioral treatment groups, the effect in first 2 weeks of treatment was a highly sensitive predictor for later stable response (80%) and stable remission (75%). Another example of a study on the timecourse of depression scores is shown in Fig. 3 [composed from [48]]. In this survey 7 randomized double-blind clinical trials of duloxetine versus placebo and comparator selective serotonin antidepressant in a total of 2515 patients with MDD were analyzed. Different trajectories of responders (77%) and non-responders (24%) were identified, whereas the placebo-treated patients showed a single response trajectory. These curves were well described as exponential decay functions; both placebo and drugresponders exhibit strikingly similar curves as described in Fig. 2

Conventionally, biological systems are considered deterministic, i.e., a brain or mood state can be predicted from a previous state. We discuss four models on the time-course of transitions in complex dynamical systems and suggest possible implications to understand mechanisms or processes allowing fast mood transitions. Firstly some approaches to model mood instability. Next three transition (T) models are discussed denoted as dynamic system critical (DSCT) [88,89], network connectivity (NCT) [90–95] and quantum mind/brain probability (QMBT) [96–100] models. Some basic features of the latter models are shown in Fig. 3. Several authors [e.g., [54]] have proposed mood instability as a core characteristic of depression. High variability of mood scores in MDD has often been compared to mood-instability in other cohorts or during previous episodes. The deviation of subsequent scores are normalized. This approach, known as approximate entropy [55], is a model-independent quantification of the regularity (complexity) of the data [54,55].The basic idea is that unpredictable (or irregular, not due to noise) variations point to multiple input mechanisms. The well-fitting stochastic (-near-exponential) models describing the time-course of MDD is consistent with this idea. The DSCT describes the interaction of two competing systems (1 and 2), where one system (1) gradually overrules the other system (2), followed by an abrupt transition and ending in recessive state of 2. Before the transition, the entire system gradually destabilizes,

J. Korf / Medical Hypotheses 82 (2014) 581–588

as observed by an increasing number of spontaneous fluctuations [56]. In line with the DSCT model. are the mood instabilities before a mood transition, as for example, the early decrease of the Hamilton scores, mood-instability or the early subclinical responses to convulsive treatment. Also the results of the experience sampling or ecological momentary assessment approach show that early sub-therapeutic responses to antidepressants anticipate later therapeutic success [43–45]. The small world network approach (NCT) emphasizes the importance of communication between the constituting elements. By introducing long-connections, the information transfer within the network becomes initially more efficient, whereas too many long-distance connections and diminution of short connections (between neighbors) might lead to disintegration [57]. Recently, a network approach to depression symptoms was described to verify the DSM-classification [58]. The basic idea is that symptoms do not merely reflect passive psychometric indicators of a latent condition, but have autonomous causal power. The network model containing all individual DSM-IV symptoms shows that symptoms of the same disorder were directly connected, together with indirect connections through shared symptoms. Four central nodes in the DSM-IV graph associated with mood and anxiety disorders were irritable, distracted, anxious and depressed symptoms. Half of the symptoms in the network were connected to depressive symptoms. The rapidly acting antidepressant ketamine has been attributed to changing brain network connectivity [59]. The application of moment-to-moment mood scoring allows the construction of personal networks underlying the individual’s psychopathology at the smallest level. Similar causes for individual does not necessarily result in similar outcomes and -vice versa – a particular diagnosis is not necessarily the consequence of common causes [Wichers [42]]. Our transition hypothesis might best be regarded as cluster property: MDD should be considered in the context of complex mutually reinforcing networks of causal mechanisms. QMBT models exclude the simultaneous existence of two alternative states and explain fast switches between brain states [60]. The QMBT model assumes that the brain confer to a special state (in quantum terms ‘superposition’) which is a prerequisite for a fast transition into another, e.g., a non-depressed, state. QMBT models have thus far been applied to decision-making, to classical ambiguity pictures and might apply to execution of voluntary movements [60,61]. Fast models of transitions are compatible with the (stochastic or random) dynamics of mood transitions seen in MDD [section time-to-recovery]. The transition models thus far explored have not been rigorously tested in MDD, but they may eventually challenge the concept of depression as a disorder. If stochastic or random transitions are indeed prominent features of MDD, depression may be considered more a symptom such as pain, rather than a disorder with a specific time course. It is too premature to decide whether and which discussed models apply to MDD.

Concluding remarks We elaborated on the idea that depression is a transient state (of the brain), rather than a disease or disorder with a predictable time course. As far as we know, the chosen approach has not extensively been reviewed previously. The concept’s supporting arguments are based on clinical and epidemiological studies. Of importance is that 1st fast mood transitions might be precipitated through a apparently unrelated mechanism, 2nd the natural course of depression is rather diverse and both early and late therapeutic responses have been reported, 3rd unpredictable and random intra- and inter-subject variations of the course of depressive

587

episodes were observed, 4th individual trajectories shows mood instabilities during the process of recovery from MDD. Among the limitations of our proposal is the lack of direct evidence: the concept was illustrated by observational rather than hypothesis-driven studies. A major limitation is the use of semiquantitative depression scores on which the exponential models are based. Hence length of a depressive episode might be determined by the change of only a few sub-scores. This objection isat least partially-circumvented by consulting a wide variety of unrelated sources (epidemiological and therapy intervention studies and case reports). The question of homogeneity of the cohorts has not thoroughly been addressed: hence part of the conclusions may be due to ‘natural’ variations of depression and heterogeneity of the patients. The latter might also be due to the current classification systems (such as the DSM) that are to a large extent ex cathedra formulated and not easily amenable to scientific falsification. [1,3,62] Whether MDD is a (theoretical) construct that does not refer to the real world is still today a fundamental (philosophical) question [62]. We might consider the possibility that (in some MDD patients) the depressive mood is composed of few small mood states (‘‘mood quanta’’), so large mood sweeps are unlikely or even impossible. On the other hand the therapeutic potential of the fast acting therapeutic interventions has as yet not fully been exploited. Real progress is possible only when both the scientific and clinical views converge to an unambiguous conceptualization, that has certainly not yet been achieved with MDD. Conflict of Interest Statement No conflict of interest. Acknowledgements The encouraging discussions with and contributions of Siebren van der Werf, Peter de Jonge, Anatolyi Gladkevich, Drozdstoy Stoyanov, Bennard Doornbos, Kirsten Kaptein, Marit Tanke, Fokko Bosker and Hans Klein are highly appreciated. My work was supported by the University of Groningen and the foundations Breinbreuk and Topsensor. Figs. 2 and 3 were composed from the data of [32–34,37]. A minor part of Fig. 4 was redrawn from the publication [58] in PloS ONE. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. References [1] Stojanov D, Korf J, de Jonge P, Popov G. The possibility of evidence-based psychiatry: depression as a case. Clin Epigenetics 2011;2(1):7–15. http:// dx.doi.org/10.1007/s13148-010-0014-2. [2] Mössner R, Mikova O, Koutsilieri E, et al. Consensus paper of the WFSBP task force on biological markers: biological markers in depression. World J Biol Psychiatry 2007;8(3):141–74. [3] Korf J, Bosker FJ. The depressed patient in a biological world: on philosophical and diagnostic strategies. J Eval Clin Pract 2013;19:514–21. [4] Cuijpers P. Effective therapies or effective mechanisms in treatment guidelines for depression? Depression Anxiety 2013;00:1–3. [5] American Psychiatric Association. Diagnostic and statistical manual of mental disorders (4th and following editions) (DSM IV). Washington DC: American Psychiatric Association; 1994. [6] Bejjani BP, Damier P, Arnulf I, et al. Transient acute depression induced by high-frequency deep-brain stimulation. N Engl J Med 1999;1340:1476–80. [7] Blomstedt P, Hariz MI, Lees A, et al. Acute severe depression induced by intraoperative stimulation of the substantia nigra: a case report. Parkinsonism Relat Disord 2008;14(3):253–6. [8] Tommasi G, Lanotte M, Albert U, et al. Transient acute depressive state induced by subthalamic region stimulation. J Neurol Sci 2008;273(1–2):135–8. [9] Mahgoub NH, Kotbi N. Acute depression and suicidal attempt following lowering the frequency of deep brain stimulation. J Neuropsychiatry Clin Neurosci 2009;21(4):468.

588

J. Korf / Medical Hypotheses 82 (2014) 581–588

[10] Kramlinger KG, Post RM. Ultrarapid and ultradian cycling in bipolar affective illness. Br J Psychiatry 1996;168:314–23. [11] Bauer M, Glenn T, Grof P, et al. Self-reported data from patients with bipolar disorder: frequency of brief depression. J Affect Disord 2007;101(1–3):227–33. [12] Strejilevich SA, Martino DJ, Murru A, Teitelbaum J, Fassi G, Marengo E, et al. Mood instability and functional recovery in bipolar disorders. Acta Psychiatr Scand 2013;128(3):194–202. http://dx.doi.org/10.1111/acps.12065. [13] Bauer M, Beaulieu S, Dunner DL, Lafer B, Kupka R. Rapid cycling bipolar disorder – diagnostic concepts. Bipolar Disord 2008;10:153–62. [14] Wirz-Justice A, Van den Hoofdakker RH. Sleep deprivation in depression: what do we know, where do we go? Biol Psychiatry 1999;46:445–53. [15] Giedke H, Schwärzler F. Therapeutic use of sleep deprivation in depression. Sleep Med Rev 2002;6:361–77. [16] Wiegand M, Riemann D, Schreiber W, et al. Effect of morning and afternoon naps on mood after total sleep deprivation in patients with major depression. Biol Psychiatry 1993;33:467–76. [17] Beck J, Hemmeter U, Brand S, et al. Modafinil reduces microsleep during partial sleep deprivation in depressed patients. J Psychiatr Res 2010;44(13):853–64. [18] Wiegand MH, Lauer CJ, Schreiber W. Patterns of response to repeated total sleep deprivations in depression. J Affect Disord 2001;64(2–3):257–60. [19] Daan S, Beersma DGM, Borbély AA. Timing of human sleep: recovery process gated by a circadian pacemaker. Am J Physiol 1984;246:R161–78. [20] Duncan Jr WC, Zarate Jr CA. Ketamine, sleep, and depression: current status and new questions. Curr Psychiatry Rep 2011;15:394–402. http://dx.doi.org/ 10.1007/s11920-013-0394-z. [21] Andrade P, Nobless LHM, Temel Y, et al. Neurostimulatory and ablative treatment options in major depressive disorder: a systematic review. Acta Neurochir 2010;152:565–77. [22] Wani A, Trevino K, Marnell P, Husain MM. Advances in brain stimulation for depression. Ann Clin Psychiatry 2013;25(3):217–24. [23] Allan CL, Ebmeier KP. The use of ECT and MST in treating depression. Int Rev Psychiatry 2011;23(5):400–12. http://dx.doi.org/10.3109/ 09540261.2011.614223. [24] Kellner CH, Greenberg RM, Murrough JW, et al. ECT in treatment-resistant depression. Am J Psychiatry 2012;169:1238–44. [25] Segman RH, Shapira B, Gorfine M, Lerer B. Onset and time course of antidepressant action: psychopharmacological implications of a controlled trial of electroconvulsive therapy. Psychopharmacol (Berl) 1995;119:440–8. [26] Hamani C, Mayberg H, Snyder B, et al. Deep brain stimulation of the subcallosal cingulate gyrus for depression: anatomical location of active contacts in clinical responders and a suggested guideline for targeting. J Neurosurg 2009;111:1209–15. [27] Aleman A. Use of repetitive transcranial magnetic stimulation for treatment in psychiatry. Clin Psychopharmacol Neurosci 2013;11(2):53–9. [28] DeBattista C, Posener JA, Kalehzan BM, Schatzberg AF. Acute antidepressant effects of intravenous hydrocortisone and CRH in depressed patients: a double-blind, placebo-controlled study. Am J Psychiatry 2000;157:1334–7. [29] Krystal JH, Sanacora G, Duman RS. Rapid acting glutamatergic antidepressants: the path to ketamine and beyond. Biol Psychiatry 2013;73:1133–41. [30] Drevets WC, Furey ML. Replication of scopolamine’s antidepressant efficacy in major depressive disorder: a randomized, placebo-controlled clinical trial. Biol Psychiatry 2010;67:432–8. [31] Cowen PJ. Serotonin and depression: pathophysiological mechanism or marketing myth? Trends Pharmacol Sci 2008;29:433–6. [32] Van der Werf SY, Kaptein KI, de Jonge P, et al. Major depressive episodes and random mood. Arch Gen Psychiatry 2006;63(5):509–18. [33] Kaptein KI, De Jonge P, Korf J, et al. Random-mood interpretation of determinants for major depression. Psychol Med 2007;37(9):1261–71. [34] de Jonge P, Conradi HJ, Kaptein KI, et al. Duration of subsequent and periods of recovery in recurrent major depression. J Affect Disord 2010;125(1–3):141–5. [35] Vuorilehto MS, Melartin TK, Isometsä ET. Course and outcome of depressive disorders in primary care: a prospective 18-month study. Psychol Med 2009;39:1697–707. [36] Keitner GI, Ryan CE, Miller IW, Norman WH. Recovery and major depression: factors associated with twelve-month outcome. Am J Psychiatry 1992;149:93–9. [37] Patten SB. A major depression prognosis calculator based on episode duration. Clin Pract Epidemiol Mental Health 2006;2(13):1–8. [38] Pezawas L, Angst J, Kasper S. Recurrent brief depression revisited. Int Rev Psychiatry 2005;17(1):63–70.

[39] Maier W, Gänsicke M, Weiffenbach O. The relationship between major and subthreshold variants of unipolar depression. J Affect Disord 1997;45(1– 2):41–51. [40] Heiby EM, Pagano IS, Blaine DD, et al. Modeling unipolar depression as a chaotic process. Psychol Assess 2003;15:426–34. [41] aan het Rot M, Hogenelst K, Schoevers RA. Mood disorders in everyday life: a systematic review of experience sampling and ecological momentary assessment studies. Clin Psychol Rev 2012;32(6):510–23. http://dx.doi.org/ 10.1016/j.cpr.2012.05.007. [42] Wichers M. The dynamic nature of depression: a new micro-level perspective of mental disorder that meets current challenges. Psychol Med 2013;14:1–12. [43] van de Leemput, IA. Wichers M, Cramer AOJ, et al. Critical slowing down as early warning for the onset and termination of depression. PNAS 2013 in press. [44] Wichers MC, Barge-Schaapveld DQ, Nicolson NA, et al. Reduced stresssensitivity or increased reward experience. The psychological mechanism of response to antidepressant medication. Neuropsychopharmacol 2009;34(4):923–31. http://dx.doi.org/10.1038/npp.2008.66. [45] Höhn P, Menne-Lothmann C, Peeters F, et al. Moment transfer of positive emotions in daily life predicts future course of depression in both general population and patient samples. PLoS One 2013;8(9):e75655. [46] Tadic´ A, Helmreich I, Mergl R, et al. Early improvement is a predictor of treatment outcome in patients with mild major, minor or subsyndromal depression. J Affect Disord 2010;120:86–93. [47] Katz MM, Meyers AL, Prakash A, et al. Early symptom change prediction of remission in depression treatment. Psychopharmacol Bull 2009;42:94–107. [48] Gueorguieva R, Mallinckrodt C, Krystal JH. Trajectories of depression severity in clinical trials of duloxetine: insights into antidepressant and placebo responses. Arch Gen Psychiatry 2011;68(12):1227–37. http://dx.doi.org/ 10.1001/archgenpsychiatry.2011.132. [49] Papakostas GI, Petersen T, Sklarsky KG, et al. Timing of clinical improvement and symptom resolution in the treatment of major depressive disorder. Psychiatry Res 2007;149:195–200. [50] Seemüller F, Riedel M, Obermeier M, et al. Outcomes of 1014 naturalistically treated in-patients with major depressive episode. Eur Neuropsychopharmacol 2010;20:346–55. [51] Uher R, Muthen B, Souery D, et al. Trajectories of change in depression severity during treatment with antidepressants. Psychol Med 2010;40(8):1367–77. [52] Gomeni R, Merlo-Pich E. Bayesian modeling and ROC analysis to predict placebo responders using clinical score measured in the initial weeks of treatment in depression trials. Br J Clin Pharmacol 2007;63:595–613. [53] Gomeni R, Lavargne A, Merlo-Pich E. Modeling placebo response in depression trials using a longitudinal model with informative dropout. Eur J Pharm Sci 2009;36:4–10. [54] Bowen RC, Wang Y, Balbuena L, et al. The relationship between mood instability and depression: implications for studying and treating depression. Med Hypotheses 2013;81(3):459–62. http://dx.doi.org/10.1016/ j.mehy.2013.06.010. [55] Pincus SM. Approximate entropy as a measure for psychiatric serial metrics. Bipolar Disord 2006;8:430–40. [56] Scheffer M, Carpenter SR, Lenton TM, et al. Anticipating critical transition. Science 2012;338(6105):344–8. http://dx.doi.org/10.1126/science.1225244. [57] Bechtel W. Network organization in health and disease: on being a reductionist and a systems biologist too. Pharmacopsychiatry 2013;46(Suppl 1):S10–21. http://dx.doi.org/10.1055/s-0033-1337922. [58] Borsboom D, Cramer AO, Schmittmann VD, et al. The small world of psychopathology. PLoS One 2011;6(11):e27407. http://dx.doi.org/10.1371/ journal.pone.0027407. [59] Scheidegger M, Walter M, Lehmann M, et al. Ketamine decreases resting state functional network connectivity in healthy subjects: implications for antidepressant drug action. PLoS One 2012;7(9):e44799. http://dx.doi.org/ 10.1371/journal.pone.0044799. [60] Bruza PD, Busemeyer JR. Quantum models of cognition and decision. Cambridge: Cambridge University Press; 2012. p. 1–403. [61] Pothos EM, Busemeyer JR. Can quantum probability provide a new direction for cognitive modeling? Behav Brain Sci 2013;36:255–327. [62] Kendler KS, Zachar P, Craver C. What kinds of things are psychiatric disorders? Psychol Med 2011;41(6):1143–50. http://dx.doi.org/10.1017/ S0033291710001844.