Abstract
In recent years, the early phase of psychosis has become an important field of research in psychiatry, and the therapeutic strategies which were developed as a consequence of this growing interest may be beginning to improve the outcome in these potentially disastrous conditions [1]. Probably partly in reaction to the pessimism with which it was considered, schizophrenia has drawn most of the attention in this domain, to the point that ‘early psychosis’ is often equated to ‘early schizophrenia’. As a consequence, first-episode (FE) affective psychoses have been relatively neglected. Mania in particular has been understudied, and the literature about FE mania is sparse. For many reasons, it is imperative that a preventive approach should be extended to the full range of FE psychosis.
First, mania is a frequent disorder, its lifetime prevalence in the USA being 1.6% [2]. Second, while the Kraepelinian view of mental illness was excessively pessimistic regarding schizophrenia, it has been excessively optimistic regarding manic-depression [3–6]. Third, many studies have shown that the number of manic episodes is a predictive factor for greater risk of relapse [7], more severe cognitive deficits [8–11] and worse overall outcome [5]. Besides the disruptive effect mood symptoms have on patient lives, up to 15% of patients can be expected to commit suicide without sustained therapy [12]. As expressed by K. Jamison [13]: ‘[This] disease that has, on several occasions, nearly killed me does kill tens of thousands of people every year: most are young, most die unnecessarily, and many are among the most imaginative and gifted that we as a society have’. It is likely that the key principles of early intervention in psychosis, namely early detection and optimal, intensive and sustained intervention during the early years of illness, apply equally well to bipolar patients as they do generally in early psychosis. As a consequence, there is a need for better knowledge of the clinical aspects of the early phase of mania, of the outcome and its predictors, of the efficacy of the various available treatments, and for the development of early intervention and possibly preventative strategies. This review is intended as a step in this direction.
In the following article, the authors summarize the principal elements relating to FE mania in the literature. The main computerized databases [Medline, PubMed, Current Contents] have been accessed utilizing the key words ‘first-episode’, ‘first-episode psychosis’, ‘mania’, ‘bipolar disorder’, and ‘schizoaffective disorder’.
Why focus on the first episode?
The study of homogeneous populations of FE patients, assessed prior to the administration of psychotropic medication, offers various advantages and avoids important confounding variables. First, while trying to identify possible aetiological factors, the study of FE patients may reduce the contamination of their effect from manifestations of the disease itself, or of possible secondary changes related to the duration of the illness [14]. Second, the identification of premorbid and prodromal factors and the study of their influence on the course of the illness is also likely to be more reliable when FE patients are studied prospectively [14]. Third, because medication has an influence on the clinical presentation and can sometimes mimic psychiatric symptoms, it is better to assess patients before or just after the beginning of the treatment [15]. Fourth, since it is known that prior episodes strongly predict future episodes of mania, the inclusion of multi-episode patients induces a negative selection bias [16–18]. Fifth, similarly to what has been described in schizophrenia by Wyatt et al. [19] and by Liebermann [20], Post proposed in 1992 the neuronal sensitization model of bipolar disorders [21]. This model suggests that the activation of neurotransmitter pathways during an affective episode produces not only acute modifications, but also more permanent alteration in the neuronal activity possibly transduced at the level of gene expression [22]. These alterations might in turn induce a higher tendency to relapse, with manic episodes being more easily triggered and of a longer duration. These changes occur early in the course of the illness, hence focusing on FE patients reduces their influence. Sixth, the study of FE patients allows a representation of the full range of possible outcome in the context of a specific illness, including patients who will prove to experience only one episode of the illness in their lifetime.
Mania: definitions
In contemporary psychiatric classifications, the term mania refers to a syndrome that can be observed in the context of various diagnoses. In the literature there is however, a lack of consistency in the way authors select the various diagnoses they include in a mania study.
When Falot and Ballinger in 19th century France first defined manic depressive diseases as a clinical entity, the term clearly referred to a clinical condition. This changed in 1895 when Kraepelin confusingly included outcome characteristics in the definition to create the dichotomy between manic-depressive psychosis and dementia praecox. In his conception, manic depressive patients would recover completely between episodes while dementia praecox would follow a progressively deteriorating path. However, the frequent association with psychotic symptoms often lead to the confusion of mania with schizophrenia and the blurring of the boundaries between the two entities has many times been pointed out. Since that time, the classification of psychiatric conditions has been a continuous matter of debate, and has grown in complexity.
In 1975 Kendell [23] suggested that more research attempting to classify psychotic illnesses as discrete types might in fact blur diagnostic distinctions, whereas considering them in dimensional terms might prove more useful. Along the same lines, McGorry et al. [24] recently conducted a factor analysis carried out on the psychopathological items of 509 FE patients that resulted in a four-factor solution (positive, negative, manic and depressive). These four factors explained 43.6% of the variance, and seemed clinically valid. The authors suggested that, in early psychosis in particular, until true aetiological or pathophysiological validators are discovered, focusing on functional psychosis rather than on discrete diagnosis such as schizophrenia, would allow for more progress both in basic research and clinical care development. Van Os et al. showed that a dimensional rather than categorical classification of psychosis also offers clinical advantages in a chronic population [25].
Although it overlaps with various diagnostic entities, mania, in terms of a particular clinical picture composed of excessive mental and physical activity, is a clinically relevant focus. Its occurrence implies certain common treatment issues, and mania certainly is one of the most challenging clinical presentations of mental illness in psychiatric practice. In this review, mania is used in a broad syndromic sense, including bipolar patients regardless of the presence or not of psychotic symptoms, as well as schizoaffective patients with manic episodes.
Outcome and FE mania
Concept and definitions of outcome
Outcome is a complex concept and its evaluation can be influenced by many factors. One of them is the way the boundaries of an illness are defined. When basing his classification mainly on outcome variables, Kraepelin sought to validate putative diagnostic entities and hoped to increase our understanding of the nature of key psychiatric conditions. However, he introduced a major selection bias, excluding by definition patients with poor outcome from any mania cohort. Similarly, many clinicians still regard patients with acute schizophrenia who recover fully as not really having had ‘true’ schizophrenia. The substitution of course and outcome as a marker for unknown biological basis of dementia praecox rigidly linked syndrome and course of the illness, and has had serious clinical implications. While this assumption has since been questioned often, in particular by Kraepelin himself in his later works, it still has a strong influence on the way schizophrenia and bipolar illnesses are considered [26], which has been detrimental to both. With the works of Bleuler and then of Schneider, the diagnostic boundaries of schizophrenia broadened, and some of the patients who would formerly have been diagnosed as manic-depressive were classified as schizophrenic. The diagnosis of manic-depressive illness was then applied in an even more restrictive manner [27]. In 1980, the DSM-III brought back a more restrictive and more descriptive definition of schizophrenia. Since then a significant decline in the rate of schizophrenia and a relative increase in the frequency of major affective disorders has been observed [27], the diagnostic category of bipolar disorders being nowadays more broadly defined. This last shift probably led to the reincorporating of patients with a worse prognosis in the bipolar cohorts. Along with the widespread failure to effectively implement lithium therapy in clinical practice, this factor might help to explain why in the last few years, many papers came to challenge the idea that the outcome of mania is generally good.
The evaluation of outcome is also influenced by the terms in which it is defined. Some authors consider it in terms of the occurrence of a relapse. However, besides the possible difficulties in defining relapse, the recurrence of symptoms gives only a partial view of the functioning of a patient. More often outcome is evaluated by assessing the patients according to various criteria at predetermined time intervals. The choice of these variables obviously has a major influence on the perception we can have of the outcome. In 1988, Dion showed in a cohort of 44 FE and multi-episode bipolar patients followed up 6 months after discharge from hospital that while 80% of them were symptom free, only 43% were employed and only 21% of them were working at their expected level of employment [3]. In other words, even if the symptomatic outcome was quite good, reflecting the efficacy of medication in controlling symptoms, the functional outcome (vocational and living situation in comparison with what it was one month prior to admission) was overall quite poor. Part of this discrepancy may be due to the sensitivity of psychosocial processes (e.g. employment, friendship, and familial relations) during adolescence and early adulthood. This also holds for non-affective psychosis and accounts for a significant proportion of failure of recovery in schizophrenia as well. Tohen et al. showed that the marked discrepancy between syndromic recovery (not meeting the criteria for an ongoing DSM IV illness episode) and functional recovery was sustained at 4-year follow up [5]. Besides, psychosocial impairment extends to all areas of functioning, persists for years, even if sustained resolution of clinical symptoms occur, and despite the use of prophylactic mood stabilizers [3–6]. Tohen et al. [15] and later Strakowski et al. [28, 29] further refined the definitions of the various domains of recovery, differentiating symptomatic recovery (8 contiguous weeks with minimal psychiatric symptoms present) from syndromic recovery [28]. They recently subdivided functional recovery into four distinct areas (role performance, interpersonal relationships, sexual activity and recreational enjoyment) that were shown to recover independently after a manic episode [29]. However, as pointed out by MacQueen et al., the studies published so far have used either too restrictive or too global measures of outcome [30]. They have therefore failed to evaluate certain domains of functioning such as, for example, a patient's sense of wellbeing and perception of their capacity to function.
Outcome of FE manic patients
As stated above, many authors have demonstrated that while the outcome of manic-depressive illness is better than that of non-affective psychosis [15, 31], it is not as favourable as was thought previously, and that restitution to the previous level of functioning is by no means the rule. This appears to be also true for FE patients albeit to a lesser extent.
However, the available literature on outcome in FE mania is sparse and hence allows only partial conclusions. As will be discussed later in this paper, numerous factors seem to influence the outcome. Differences in outcome among the various studies might actually simply reflect the differences in the study populations. In fact, most of the studies involving FE manic patients have attempted to study the influence of one or a few of these factors. This approach gives a very limited view of the problem, and fails to grasp the complexity of this illness. On the other hand, studies addressing outcome in more general terms must also be interpreted carefully. Very few studies have focused specifically on manic patients who are more often included as part of broader diagnostic groups (bipolar disorders, FE bipolar disorder, FE psychosis, FE affective psychosis, FE major affective disorder with psychotic features). The results relating to manic patients in these cohorts are frequently not specifically expressed and therefore difficult, if not impossible, to extrapolate.
In a cohort of 44 bipolar patients, Dion et al. showed that, 6 months after hospitalization for mania, 36% of the 14 FE patients of their cohort were unemployed even if they exhibited significant symptom resolution at that time [3]. In the context of the McLean-Harvard firstepisode project commenced in 1989, Tohen et al. published various studies concerning FE manic patients, in a highly selected private hospital setting. In 1990 [14] they showed that 20% of a cohort of 24 FE manic patients had relapsed after 6 months and 54% of them after 4 years. However, their functional outcome appeared to be relatively good and to improve over time: 79% at 6 months and 87% at 4 years were able to work competitively, and 83% at 6 months and 92% at 4 years lived independently. These results are, however, in contrast with later reports. In 1992, looking at the 6-month outcome of the bipolar subgroup of a cohort of 102 FE psychotic patients, they observed that, although 85% had recovered syndromically, only 68% had recovered functionally [15]. In a recent publication they showed that of the 98 FE bipolar manic patients included in a cohort of 257 FE psychotic patients, while 86% had experienced syndromal recovery at 6 months, 33% only had recovered functionally [31]. Finally, in a cohort composed of 146 pure and mixed manic FE psychotic patients, they observed that at 24 months 98% had experienced syndromal recovery but only 40% functionally [32]. In a cohort of 109 FE patients with affective psychosis, Strakowski et al. found that while 61% of the bipolar subgroup (manic, mixed and depressive) recovered sydromically after 12 months, only 36% recovered functionally [28]. They observed that recovery proceeds through stages: syndromic recovery came first and was correlated with compliance. Symptomatic recovery happened later, suggesting that additional interventions were necessary to help the patients reach this stage, and functional recovery occurred later still. In a follow up of 42 FE manic patients over 8 months, the same group showed that various areas of function (role performance, interpersonal relationships, sexual activity and recreational enjoyment) recovered independently [29]. Good functional outcome, defined as recovery of at least three of the four areas of function, was reached by 48% of the patients, while only 5% recovered all four aspects during follow up. Functional recovery did not seem to be correlated with symptomatic recovery, which was attained by 60% of the patients.
With the limitations mentioned above, these results show that the outcome of a first manic episode is not the complete recovery that Kraepelin anticipated, and shows significant overlap with the outcome of FE schizophrenia patients. While the symptoms of the acute episode remit in the majority of the cases, an important number of patients do not recover functionally. Moreover, the concept of recovery, while better defined, has so far been addressed in a too simplistic manner and has failed to evaluate important aspects of functioning. If the effect of lost opportunities and involvement with peer group, as well as subjective wellbeing were to be taken into account, outcome of FE mania might prove even worse than it has been described in the literature so far.
Comparison of the outcome of FE versus multiepisode manic patients
When comparing first admission with multiple admission patients, Dion observed that the latter had less favourable outcomes on all follow-up measures of functioning [3]. While 64% of the FE patients were employed at some level at follow up, only 33% of multiepisode patients were, and only 10% at their expected levels; 86% of the FE patients lived independently while only 56% of the multi-episode did so. In 1995, Keck et al. studied the short-term outcome (at time of discharge) of 34 FE and 37 multi-episode manic patients, and observed that multi-episode patients needed significantly longer hospitalizations to achieve similar levels of clinical improvement [22]. As suggested by Tohen et al. inclusion of a mixture of FE and multi-episodes in studies might therefore give a false impression of chronicity to the course of manic patients [14].
Comparison of mixed and pure manic patients
The term ‘mixed episode’ refers to the state of a patient who rapidly alternates between manic and depressive mood. McElroy et al. [33], and Keller et al. [34, 35] reported that pure manic patients had a better outcome than those with mixed episode. In contrast, Keck et al. [36] found no difference on any outcome variable between both groups. In FE patients, the few results available are conflicting. Tohen et al. [37] showed that the mixed mania group had a worse outcome at the 2-year follow up, only 33% of them being able to live independently compared to 82% of the pure mania group. However, they also found that comorbidity was strongly associated with mixed episodes. In contrast, Strakowski et al. found in their cohort that 52% of mixed compared with 28% of the pure manic patients recovered functionally [28].
Predictors of outcome in FE mania
While it is thought that identification of premorbid baseline factors that may predict the future course of an illness is likely to be more valid if pure samples of FE patients are studied prospectively [14], such studies are not numerous. In the others, the proportion of FE cases varies from study to study, which probably explains the discrepancies in their results.
Socio-economic variables
In cohorts of multi-episode patients, Tohen et al. have shown that poor occupational status prior to the index episode was an unfavourable predictor of outcome [5]. Similarly, Coryell et al. showed that good friendship pattern during adolescence and high maximum level of functioning during the 5 years prior to the manic episode were predictive of good recovery [38]. Keck et al. observed in a cohort of 134 multi-episode bipolar patients that better functional recovery was associated with higher social class [36]; however, with chronic patients it is hard to know if the poor social adjustment is a consequence of prior episodes, hospitalization and the disruption they induce in social life [29].
However, among FE patients, the results are similar. In a cohort of 24 FE manic patients, Tohen et al. showed that low occupational status at baseline predicted poor global social adjustment at 4 years [14]. The evaluation of the degree of recuperation of the previous level of functioning, rather than the social adjustment itself, is however, a better way to evaluate functional outcome. Strakowski et al. observed that higher socioeconomic status was correlated with both better symptomatic and functional recovery, and good premorbid functioning with better functional recovery [28]. In a subsequent study, they observed that premorbid socioeconomic status in terms of education and employment was significantly correlated with recovery of role performances and overall good outcome [29]. Tohen et al. in a cohort of 219 FE major affective disorders showed that time to syndromal recovery was significantly shorter for patients who were married [32].
Age at onset of the illness
Mania appears to present differently when it develops in adolescence, with patients showing more comorbidity, neurodevelopmental and premorbid psychosocial difficulties [39]. They also appear to have higher rates of psychotic symptoms than patients developing the illness later in life [40]. This might be partially related to a physiological and neurodevelopmental stage through which the brain normally goes during late adolescence and early adulthood, which favours the clinical expression of psychotic symptoms [41]. Sax et al. found that early onset patients (before age 18) had more energy, minimal sleep disruption, and greater suicidality [42]. Schürhoff et al. recently collected data suggesting that early and late onset bipolar disorders may represent different forms of manic-depressive illness with different clinical expression and familial risk [43]. These differences in presentation might explain why bipolar illness is often unrecognized or misdiagnosed in adolescence [40]. Joyce showed that patients with an early onset of bipolar affective disorder, especially if they were first hospitalized for mania, were most likely to have received a diagnosis of schizophrenia [44]. The subject is, however, made more complex by inconsistencies in defining date of onset (first symptoms, first episode, first treatment or first hospitalization) [45]. Despite these difficulties, these studies have reported some interesting results.
Carlson et al. studied a cohort of 23 early (before age 21) and 30 late (after age 30) onset psychotic manic patients over 24 months, an observation period which permitted stabilization of the diagnosis [39]. The results showed that early onset patients were significantly more likely to be male, to have had childhood behaviour disorder, to have substance abuse comorbidity and to exhibit grandiose and paranoid delusions. They were less likely to experience complete episode remission than the adult onset patients. Tohen et al. also reported a better functional recovery by 2 years for those of his 219 FE major affective disorder patients who were older (more than 30 years of age) at the onset of the illness [32]. In a retrospective study, Johnson et al. explored the possible correlation between hereditary and psychosocial factors and the age of onset of bipolar and unipolar affective disorders [46]. They showed that bipolar patients with high constitutional vulnerability (family history of affective disorder) were prone to an earlier age of onset and needed fewer stress factors to become ill.
These results suggest that a subtype of patients with hereditary loading have an earlier age of onset, need a lower degree of stress before they develop the illness, tend to have a worse outcome and present with more comorbid diagnoses. However, there are two other ways to interpret these data. First, the possible ‘toxic’ effect of episodes suggested by Post might have more impact on a younger and more immature brain. Second, due to the sensitivity of psychosocial processes during adolescence and early adulthood, a first episode developing at a younger age might have more deleterious effects than if it occurs when the patient has had a chance to develop social skills and a social network.
Presence of psychotic features
Psychotic symptoms often occur during manic episodes. Goodwin and Jamison in a review of 26 studies of psychotic features in mania concluded that close to 60% of patients with a bipolar disorder had a lifetime history of at least one psychotic symptom [47]. The prevalence of psychotic symptoms among FE manic patients has not been studied specifically. Estimates from the available published data suggest they are present in 63–88% of cases [14, 29, 40]. Despite the views of Kraepelin, Bleuler and especially Kurt Schneider who asserted that some psychotic symptoms were more typical of schizophrenia and others more typical of mania, it has been shown that all types of psychotic symptoms can occur in mania [48]. Mood congruent psychotic symptoms are probably the most common, but mood-incongruent psychotic symptoms are also frequent. Fennig et al. [49] showed in a group of 49 FE psychotic bipolar patients that 20.4% of them had mood incongruent psychotic symptoms only and that 55.1% had both mood incongruent and mood congruent psychotic symptoms.
The influence on outcome of the presence of psychotic symptoms during a manic episode remains controversial. In 1983, Rosen et al. showed in a cohort of 89 multiepisode bipolar I patients that those who presented psychotic symptoms at some point of their illness had a significantly poorer outcome in terms of social functioning [50]. Tohen et al., in a 4-year follow up of 24 FE manic patients, found that the presence of psychotic features during the manic episode was significantly correlated with shorter time in remission [14]. Albus et al. showed that FE affective patients with psychotic symptoms had worse cognitive functioning compared to those without, the latter performing as well as healthy controls [51]; however, other authors [4, 14, 28, 30] failed to show any correlation between outcome and the presence of psychotic symptoms.
More specifically, the influence of the mood congruence or incongruence of the psychotic symptoms on the outcome has been studied, some cohorts including FE patients. Reflecting the fuzziness of the diagnostic boundaries between bipolar illness and schizophrenia, the diagnostic classification of patients presenting with mood-incongruent psychotic symptoms has been long debated. Many otherwise typical manic bipolar patients have been misdiagnosed with schizophrenia [48]. However, while the DSM-IV identifies a distinct subcategory for psychotic affective patients with mood incongruent symptoms, it does not provide explicit guidelines on how to group patients with both congruent and incongruent symptoms. This leads to different uses of the concept in the studies and difficulties in the classification of patients who present with both type of symptoms [52]. Tohen et al. studied a cohort of 54 patients among which 30% presented with a first episode of mania [53]. They found that the patients with mood incongruent psychotic features during the index manic episode had had a significantly shorter time in remission at 4 years and that the presence of Schneiderian first-rank symptoms was a predictor of poor residential status at 4 years; however, the inclusion of 70% of multi-episode patients means that the risk of relapse might also have been related to the chronicity of the illness.
Strakowski et al. recently suggested that the worse outcome they observed in FE patients presenting with mood incongruent symptoms might be linked to the persistence of psychotic symptoms after the acute manic episode [54]. Fennig et al., in a cohort of 49 FE psychotic affective patients, concluded that while the presence of mood incongruent psychotic symptoms was frequent, it was a weak predictor of outcome and questioned its prognostic utility [49]. In a consecutive review of the literature, the same authors came to the conclusion that such a categorization did not allow the identification of distinct subgroups of affective patients and suggested that the concept should either be abandoned or better defined [52].
While the data concerning the influence of psychotic symptoms occurring during a manic episode are inconsistent, the presence of psychotic symptoms separated temporally from the manic episode, that is to say a diagnosis of schizoaffective disorder of bipolar type, seem to clearly imply a worse prognosis. Comparing 12 ‘pure’ manic patients with 11 manic schizoaffective patients, all of ‘relatively recent onset’, Miklowitz et al. showed in a 9-month outpatient follow up that the latter had greater level of thought disturbances, worse level of social adjustment and tended to be less adherent with medication [55]. Strakowski et al. compared the outcome of 27 FE schizoaffective patients with 27 FE bipolar patients matched for sex, age, socioeconomic status and race [56]. At 12-month follow up, the schizoaffective patients were significantly less likely to have recovered syndromically.
Adherence
Non-adherence to treatment is known to be an important problem in any medical or psychiatric condition and has been identified as one of the major risk factors of relapse in bipolar disorders [57]. Bosco et al. showed that the rate of non-adherence with mood stabilizers was close to 50% [58]. Keck et al. showed in a cohort of 101 patients hospitalized for acute mania that 64% of them were non-adherent to their medication in the month prior to admission [59]. They also found that treatment adherence was associated with higher rates and more rapid recovery [36]. However, adherence to the medication was inversely correlated with the presence of comorbid substance related disorder, which might also have influenced the outcome.
In FE patients, the rate of non-adherence seems to be even higher, reported by some authors to be as high as 57% [60]. Among the 83 bipolar of a cohort of 109 FE affective patients, Strakowski et al. reported that 41% were fully adherent [28], 26% partially and 33% totally non-adherent. These numbers might be explained by various factors. First of all, denial is an associated feature of the illness. Besides, adjustment to the lifestyle change involved in prophylactic medication is a major challenge, especially for young people.
The impact of non-adherence on outcome of FE mania has not been well studied. Strakowski et al. found that syndromic recovery was more likely to occur for patients with full adherence than for those with partial adherence [28]. The impact of total non-adherence is more difficult to assess, non-adherent patients tending to drop out of the studies altogether. Moreover, non-adherence rarely occurs in isolation and is often combined with other poor prognostic factors such as substance abuse, and might then have an indirect as well as a direct effect on outcome. Poor adherence is a major contribution to the difference between the proven efficacy of lithium in research settings, and its effectiveness in clinical practice. This efficacy-effectiveness gap has led some researchers in the bipolar field to question the actual efficacy of lithium. In fact, its efficacy remains unsurpassed but its effectiveness in clinical practice needs to be substantially improved.
Comorbidity
Prevalence of comorbidity in FE manic patients
Recent publications have shown that the prevalence of comorbidity in manic patients is high. The fact that it is frequently already present at the first admission suggests that it is not only a consequence of long-term illness [61]. It could also either be a risk-factor for onset, an attempt to self-medicate prior to effective treatment in the case of substance abuse, or an independent illness. Besides, FE psychosis is an evolving situation, and the task of predicting which of the syndromes will become more dominant or severe is a real challenge. According to the available publications, the prevalence of psychiatric comorbidity at FE psychosis is high, around 69% [62], and ranges from 35 to 71% in FE bipolar patients [22, 37, 62–64]. The results of a study from Tohen et al. suggest that this proportion is higher for mixed FE (60%) than for pure manic episodes (27.2%) [37]. Contrary to their hypothesis, Keck et al. did not observe any difference in the rate of comorbidity between FE and multi-episode patients [22].
According to the studies mentioned above [22, 63–65] 25–56% of the patients present with multi-psychiatric comorbidity. The two most frequent psychiatric comorbid diagnosis are alcohol abuse or dependence, diagnosed in 24–39% of the patients, and drug abuse or dependence, diagnosed in 16–35% of the patients. Other comorbid psychiatric diagnoses reported are Obsessive– Compulsive disorder (OCD) (8–13%), anxiety disorders (15–18%), impulse control disorder (23–24%), Posttraumatic stress disorder (PTSD) (1.7–18%), panic disorder (6.6–8%) and phobia (4–6.6%), all of them occurring more frequently than in the general population. Some of these figures are similar to what has been shown by Bermanzohn et al. in a population of 37 chronic schizophrenic patients [64]. Forty eight point six per cent had one or more comorbid diagnosis, 27% major depression, 29.7% OCD and 10.8% panic disorder.
Comorbidity and outcome
The study of the correlation between presence of a comorbid diagnosis and outcome has yielded contradictory results. With 25–56% of the manic patients presenting multiple comorbidity [22, 62–64], the specific influence of any comorbid syndrome is difficult to establish. However, while Strakowski et al., in a study of 41 FE manic and mixed state patients, failed to show any association between outcome measures and presence of comorbidity [64], most of the studies seem to show that the presence of comorbidity negatively influences the outcome of mania. Tohen et al. observed that, in the affective subgroup of a cohort of 102 FE psychotic patients, only 35% of the patients with psychiatric comorbidity were syndromically recovered from mania at discharge compared to 61% of those without psychiatric comorbidity [15]. In another study they showed that time to syndromal recovery was shorter for patients who lacked comorbidity in a cohort of 219 FE major affective disorders [32]. Strakowski et al. found in the affective subgroup of 102 FE psychotic patients that those presenting with psychiatric comorbidity had a poorer outcome, their recovery rate being lower at discharge [63]. More specifically, Tohen et al. showed in a study of 123 FE manic patients that presence of drug abuse or dependence was related to a poor psychosocial outcome and a low probability of recovery by 2 years [37]. Similarly, Strakowski et al. observed that substance abuse predicted a lower rate or delayed onset of recovery in a cohort of 109 FE affective psychotic patients followedup at 12 months [28], probably both by a direct effect and by its negative influence on compliance. Tohen et al. showed that patients with a history of alcoholism had a greater risk of relapse, only 50% of them remaining asymptomatic at 6 months, against 78% of those without such a history [14].
Relationship between comorbid diagnosis and mania
It is also of interest to determine whether manic patients present with real comorbid diseases that have an independent evolution relative to the mania, or if the symptoms are intrinsically related to mania, occurring as a variant in the expression of the manic syndrome. A way to study this issue has been to focus upon the chronology and the course of all the co-occurring diagnoses, and the time of their respective onset. Comorbid illnesses are considered to have more chance of being independent from each other if there is a longer interval between their respective onset (usually more than 12 months), and if their clinical course varies independently. The most frequently occurring comorbid syndromes have been studied to clarify their degree of independence from mania itself and their impact on outcome variables.
Alcohol abuse or dependence is regularly reported as being the most common comorbid diagnosis in manic patients [14, 61–64, 66–68]. It is more frequent among manic patients than in the general population [64], in affective than in non-affective psychosis [62] and in mixed manic episodes than in pure manic episodes [37, 63]. First episode manic patients with a diagnosis of alcohol abuse have a poorer initial outcome [63], lower rates and delayed onset of recovery [28] and a shorter time in remission [14]. According to Kessing [68], concurrent alcoholism seems to raise the risk of recurrence of affective episodes only early in the course of the disorder, but to have no effect on recurrence in the later phases of the illness.
The study of the chronology of occurrence of alcohol abuse and mania has yielded confounding results and the degree of interdependence of both diagnoses when cooccurring is consequently unclear. In 1992, in a cohort of 41 FE manic patients, Strakowski et al. observed that the comorbidity of alcohol abuse or dependence that was present in 36% of patients had its onset more than one year before the onset of the manic syndrome [64]. However, the lack of data concerning possible prodromal symptoms prevented them from being able to exclude that alcohol was used as self medication, rather than being an independent illness. In an attempt to clarify this matter, the same group studied the course as well as the chronology of emergence of comorbid alcohol abuse [66]. In a cohort of 77 FE bipolar patients, the fact that the onset of substance abuse occurred more than one year before the first symptoms of bipolar illness in 70% of cases, and that both illnesses had independent courses, rarely cycling together, suggested that they were independent diseases. However, further prospective studies are required to clarify the exact nature of the connection between these two diagnoses.
In contrast to its possible self-medication function, alcohol has been suggested to actually induce bipolar disorders. In a cohort of 59 FE manic patients, Strakowski et al. [61] observed that patients with antecedent alcohol abuse had an older age of onset of mania. They suggested, following Winokur et al., that substance abuse may constitute a risk factor for patients who otherwise would have had a low risk for developing a bipolar illness [69]. In 1999, DelBello et al. provided support for this hypothesis [67]. In a group of 51 FE manic patients they observed that those with a history of alcohol abuse had a lower familial rate of affective illness than those without. In other words, the manic patients with alcoholic comorbidity were subjects with low genetic risk, who possibly would not have developed the illness without this additional risk factor being present. Another design to clarify this issue would be to take a sample of cases with a family history of bipolar illness and to see whether or not those with comorbid alcohol use had an earlier age of onset.
Drug abuse is also more frequently diagnosed in FE manic patients than in the general population [64]. Most of this is polysubstance abuse, practically always including marijuana [67]. Most studies show that its presence implies a worse prognosis [63], lowers the probability of recovery at 1 [28] and 2 years [37] and shortens the delay between the onset of illness and first hospitalization [61].
However, drug abuse does not seem to constitute a risk factor for developing the illness in the sense alcohol does since it is not more frequent in patients without familial history of bipolar illness. It appears more to have an aggravating effect on the disorder, justifying an early, specific and ongoing therapeutic approach since it has been shown that the majority of manic patients with a history of drug abuse resume their consumption, often within the first month after discharge [66].
Obsessive–compulsive disorder (OCD) also occurs more frequently in manic patients than in the general population. The onset of OCD occurs usually less than a year before the onset of the bipolar illness and its course parallels that of the bipolar disorder. Taking this into account, Strakowski et al. suggested that OCD symptoms in conjunction with mania could reflect a variant of bipolar disorder, more than a separate disease entity [64].
Posttraumatic stress disorder (PTSD) on the contrary appears to be a distinct illness entity. Its higher occurrence in manic patients could be explained by shared risk factors. Since most cases of PTSD were antecedent to the bipolar illness in their patient population, Strakowski et al. suggested that its occurrence might constitute a risk factor for people already at risk to develop mania [64]. Sub-threshold symptoms of mania might also influence the behaviour of patients and lead them to expose themselves to potentially traumatic life events.
Cognition and FE mania
Various types of cognitive deficits have been described in bipolar illness, but the data have varied in consistency, and the field remains confused [70]. Furthermore, few studies have focused on manic patients, and bipolar or schizoaffective patients have mostly been included as control groups in studies where the main focus was on schizophrenic patients.
Results usually show that during an acute episode, the manic patients present with a diffuse pattern of cognitive impairment [71, 72]. According to some authors, these deficits are related to the clinical state, and patients fully recover between episodes. Others argue that some cognitive impairment persists in approximately one third of bipolar patients when in a euthymic state [9, 73–77]. Moreover, Gourovitch et al. studying monozygotic twins either concordant or discordant for bipolar disorder collected data suggesting that while some visuospatial deficits and verbal memory deficits may be related to disease parameters, mild attenuation in overall memory or retrieval function may be related to genetic factors associated with the illness, since they could also be observed in the healthy twin [70].
Due to the variety of cognitive tests used in the different studies, and the restricted domains of cognition addressed in others, it is very difficult to get a precise idea of the cognitive deficits of manic patients. Besides, unclear remission criteria, diagnostic heterogeneity, small sample sizes, and absence of longitudinal assessment make comparison of the various studies difficult. The inclusion in most of them of multi-episode patients also introduces at least two major confounding variables, since it has been shown that the degree of cognitive dysfunction could be related to the number of affective episodes [8–11] and could be influenced by medication [73, 78–79].
Few studies explored cognition in FE mania. In 1991, Nuechterlein et al. studied the information processing anomalies in the early stages of 25 schizophrenic and 7 bipolar manic patients, and compared them with 26 normal subjects [72]. In the forced-choice span of apprehension task, bipolar patients showed lower detection accuracy than normal controls, but returned to normal when in remission. The continuous performance test (CPT) revealed the same pattern, with significant impairment during the acute manic phase, and return to normal functioning when in remission. In 1995, Sax et al. explored the differences in performance on the CPT between pure manic and mixed manic patients at the time of admission for an acute episode [80]. Both groups performed worse than the healthy controls, with mixed manic patients showing poorer performances. In 1998 the same authors administered the computerized version of the degraded stimulus CPT shortly after admission and 2 months after discharge to a cohort of 27 FE psychotic patients (8 major depressive patients, 3 schizoaffective bipolar patients and 16 bipolar psychotic patients) [81]. The results suggest that manic patients present with attention dysfunction that is state dependent, the deficits observed on the CPT at baseline having significantly improved two months after discharge. This improvement appeared to be significantly correlated with the changes in the total score on the youth mania rating scale (YMRS) [82]. Albus et al. showed in a cohort of first episodes that affective patients (either depressive or bipolar) without psychotic symptoms did not perform significantly differently to controls [51]. Those with such symptoms performed, however, as poorly as schizophrenic patients, with the most pronounced impairment in the area of visual motor processing and attention. Mojtabai et al. on the contrary recently reported that FE bipolar psychotic patients performed significantly better than FE schizophrenic patients on a battery of neuropsychological tests [83]. All the patients were, however, only tested 2 years after the index episode, and schizophrenic patients had higher levels of positive and negative symptoms and were more likely to be on medication at the time of the tests.
Medication and FE mania
To date there are no specific guidelines for the treatment of FE mania, and mood stabilizers are the cornerstone of the treatment of the acute as well as the maintenance phase, as they are in later stages of bipolar illness [84]. There is, however, some confusion between their indication for acute and prophylactic treatment. It is often believed that prescription of lithium should be restricted to prophylaxis and therefore should be avoided in single episode cases and therefore in FE patients. Tohen et al. looked retrospectively at the prescribing practices in the McLean/Harvard first-episode mania project between 1989 and 1996 [37]. They observed that while lithium carbonate was the first line treatment in the late 1980s, by the mid 1990s divalproex sodium had become the most commonly prescribed mood stabilizer and that patients receiving this drug had a shorter length of hospital stay. While the naturalistic setting of this study did not allow firm conclusions to be drawn, the faster onset of action of this drug compared to lithium, and its better efficacy in mixed episodes, may be responsible.
Despite the availability of various mood-stabilizing agents and their proven efficacy in acute and maintenance phases, and despite the availability of benzodiazepines to help control the agitation in the acute phase [85], antipsychotics remain the most commonly prescribed adjunctive treatment for mania [48]. Extensive use of antipsychotics has been reported in the acute phase [86, 87]. Moreover, 6 months after hospitalization for a manic episode, 68% to 95% of patients who have been put on an antipsychotic medication still receive it [87, 89]. In the maintenance phase, proportions as high as 67% of patients have been reported to receive antipsychotic medication [89, 90]. This is a reason for concern for various reasons. Bipolar patients are known to be very susceptible to tardive dyskinesia [91]. Moreover, if antipsychotics rather than benzodiazepines are used for acute behavioural control, they tend to be prescribed at high dosages that can induce extra–pyramidal syndromes and lead to prolonged alienation of patients from treatment.
One recent publication by Zarate et al. assessed in a naturalistic setting the use of antipsychotics during and following the inpatient treatment of 129 FE manic patients [84]. They found that 77% of them received antipsychotic medication at discharge, and 25% at 6-month follow up. When comparing them with 69 non-affective FE psychotic patients, they showed that at both time points, while manic patients received significantly lower dosages of antipsychotic, their probability of receiving such a medication was not significantly different to the non-affective patients. Among manic patients, those who had not recovered symptomatically tended to have higher doses of medication, showing the reluctance of clinicians to reduce the dose until manic symptoms disappear. The usual significant correlation between the presence of certain factors (male gender, early age of onset, low educational level, severity of manic symptoms, history of psychotic symptoms) and the maintenance of antipsychotic treatment in bipolar patients was not found in this study. In particular, non-psychotic manic patients were as likely to receive antipsychotic medication as those who were psychotic, which the authors thought to be possibly related to the delay in onset of action of lithium. These results also suggest that the practice of antipsychotic prescription in FE manic patients is sometimes more based on habituation and tradition among the prescribers than on precise clinical criteria.
Neuroimaging and FE mania
Neuroimaging technologies have been increasingly applied to the study of psychiatric disorders. While most studies have focused on schizophrenia, affective psychoses have also been studied. The most consistent findings, as summarized by Stoll et al. in a recent review article are the following: increased white matter hyperintensities, cerebellar structural abnormalities, mid sulcal prominence and ventricular enlargement [92]. Functional neuroimaging reveals increased cholinecontaining compounds in the basal ganglia when subjects are symptomatic, and decreased activity of the prefrontal cortex when subjects are depressed.
Very few studies were performed in FE manic populations. In 1993 Strakowski et al. [93] compared the brain structure on MRI of 17 FE manic patients with controls. They showed that the patients had larger third and lateral ventricular volume and possible differences in the distribution of grey and white matter. The presence of such neuroanatomical abnormalities in the early course of bipolar disorders led the authors to consider that they might be in relation to the aetiology of the disorder. Subsequently, the same group performed a volumetric study of brain structures associated with brain pathways thought to modulate human mood states [94]. In first as well as multi-episode bipolar patients, they found a significant enlargement of the amygdala, and possibly of the thalamus and globus pallidus. The study also confirmed the presence of enlarged lateral ventricules. DelBello et al. conducted a pilot study of the MRI structure of the cerebellum of bipolar patients [96]. They compared the volumes of right and left cerebellar hemisphere and vermal regions V1, V2 and V3 in 16 FE patients, 14 multi-episode patients and 15 normal subjects. Multi-episode patients showed significant vermal V3 area volume decrease, which was not present in FE patients, suggesting its possible degenerative nature. Hirayasu et al. showed, in a cohort of 24 FE affective psychotic patients (21 of whom were manic) that the volume of the left subgenual cingulate volume of those with a family history of affective illness was significantly decreased compared to controls and to FE schizophrenic patients [96]. Finally, Velakoulis et al. showed a smaller left hippocampal volume in affective as well as in non-affective FE psychotic patients compared with controls [97].
To date, these findings cannot be explained by any single physiopathologic mechanism [92]. There are suggestions that frontal lobe, limbic, and basal ganglia circuits are implicated in maintaining normal moods and emotions, and that bipolar disorders in particular might be characterized by disturbances in the anatomy and function of these circuits [98]. The presence of anatomical lesions in FE patients is, however, congruent with the neurodevelopmental theory of schizophrenia [97], and may suggest a similar process for affective psychosis.
Discussion
It has been an enduring psychiatric myth that patients are very likely to recover completely after a manic episode. An important body of recent literature has shown that it is not so for a substantial proportion of patients when outcome is more accurately defined and more thoroughly assessed. These studies show that factors other than the duration of the illness and the number of relapses play a role in this since the majority of patients fail to recover functionally, and a significant minority syndromically, after a first episode of mania.
The severity of the presenting psychopathology of the manic episode does not appear to have much prognostic value. Strakowski et al. did not find any correlation between any of the psychopathological scores and the outcome [28], and Tohen et al. found only a statistically non-significant correlation between outcome and the severity of the episode at baseline [32]. While controversial, the literature reviewed suggests that the presence of psychotic symptoms and their nature in terms of congruence to the mood do not have a strong predictive value either. Furthermore, the suggestion that patients presenting with a mixed episode might do worse has not been clearly verified in FE patients and the trend observed in some studies seems to be related to the highest prevalence of comorbid diagnosis in this subgroup rather than to anything else. The only exception appears to be the occurrence of psychotic symptoms outside of a manic episode, in which case a diagnosis of schizo-affective disorder can be made, which clearly implies a poorer outcome.
Various predictors of outcome have, however, been identified. The proven influence of the socio-economic status suggests that a better original social integration allows for a faster recovery. It also explains some of the discrepancies between studies that include patients of different social background. This suggests that any intervention that could delay the occurrence of the first episode, limit the rate of relapse during the first years of illness and enable the patient to develop and secure his social network, might have a protective impact on outcome. The better prognosis of patients developing the illness later also supports this hypothesis. Conversely, however, this may be related to a lower sensitivity of the brain to the toxic effect of the episodes suggested by Post [21], or to the fact that early onset bipolar illness might constitute a distinct illness with a stronger biological base. More research is needed to clarify these aspects. The presence of psychiatric comorbidity is another important prognostic factor. Substance abuse, the most common comorbid feature, appears to induce episodes, mainly in the case of alcohol abuse, and to have an overall aggravating effect on the illness in the case of drug abuse. However, since the various comorbidities often occur together and interact with one another, their relative impact on outcome is difficult to measure. The patients should therefore be assessed thoroughly in this regard and comorbid diagnoses should be addressed in parallel with the basic treatment of the manic episode. Finally, adherence to the treatment is a crucial influence on outcome, and the development of facilitating strategies, such as adherence therapy modules and more specialized clinics, should urgently be considered.
Diagnosis in FE psychosis presents some difficulties. It has been shown that between the first and the second episode, the diagnosis changes in up to 50% of patients and that the misclassification rate during the first episode is as high as 35%, depending on the diagnostic procedure employed [26]. Therefore, the classification proposed in the DSM-IV might be inappropriate for this phase of the illnesses. Moreover, most of the publications have so far dwelled on the differences between affective and non-affective psychosis. We are struck by the many similarities between the two conditions and the degree of flux and overlap. Poor functional outcome in an important proportion of patients following the first episode, high prevalence of comorbid diagnoses, worse outcome with a younger age at onset, early presence of neuroanatomical changes and the possibility of deleterious consequences of repeated acute episodes on the structure of the central nervous system have been observed in both. A new diagnostic approach that permits a blend of dimensional and broad categorical models needs to be developed and evaluated, since a broad diagnostic set of functional psychosis would be more appropriate in this phase of the illness [24]. The outcome of FE manic patients seems to correlate better with the presence of various risk factors, which combine their effect, than with diagnostic subclassifications that attempt to group them into discrete categories. Future research should therefore aim at defining these risk factors more accurately, and at evaluating their relative impact and their degree of interdependence, in order to allow for the development of specific and targeted therapeutic strategies.
These similarities also suggest that FE manic patients might benefit from the progress made in the treatment of FE schizophrenic patients. Studies conducted among FE patients have shown that the prognosis of mania can often be as poor early after the first episode, and therefore does not depend only on the number of relapses. However, it may be that a reduction of the duration of untreated mania, similarly to the diminution of the duration of untreated psychosis, might impact on the risk of relapse and therefore on outcome [99, 100]. In 1992, Post highlighted the critical importance to implement early intervention strategies for recurrent affective disorders in order to prevent malignant transformation to rapid cycling, spontaneous episodes and refractoriness to treatment [21]. The promising methods of the few pilot studies conducted among prodromal psychotic patients [101–103] could be applied more specifically to prodromal manic patients with the goal of developing specific early intervention strategies. However, such early interventions require a good definition of the early signs of an illness with features that have a high positive predictive value. The knowledge we have about warning signs of an affective episode in bipolar patients comes currently exclusively from research conducted with chronic patients and is relative to the period preceding the occurrence of a relapse. New research is therefore needed to define prodromal symptoms of mania, to be able to select a population at high risk of developing the illness and to develop primary preventative strategies. A retrospective study that focuses on the time directly preceding the first diagnosable bipolar affective disorder episode, similar to what has been done to define the prodrome of psychosis [101], is needed as a preliminary step in this direction. Recent publications by Manji et al. demonstrating the neuroprotective effects of lithium highlight the possibility of more specific preventive treatments in this group of patients [104]. These developments have the potential to limit the tragically high rate of suicide among manic patients, and to influence the outcome of this illness, so that the optimism with which FE mania has so far been considered might ultimately become much more justified.
Footnotes
Acknowledgements
This study was supported by scholarships to Philippe Conus from the Société Académique Vaudoise, the Fondation du 450ème Anniversaire de l'Université, the Service des Affaires Universitaires and the Departement Universitaire de Psychiatrie Adulte of Lausanne, Switzerland.