1.
Introduction
guidelines for the biological treatment of bipolar disorders (Grunze et al. 2002a; Grunze et al. 2003) concerned the acute treatment of
mania and bipolar depression. Although it is of great importance to control these acute manifestations of illness as rapidly and
effectively as possible, the real key to treatment of bipolar disorder is successful maintenance treatment. Starting with Kraepelin (1921),
several long-term observational studies have demonstrated that the duration of the symptom-free interval tends to decrease sharply
with a growing number of episodes (Zis et al. 1980; Angst 1981; Roy-Byrne et al. 1985; Kessing 1998). The central aims of long-term treatment
are the prevention of new clinically significant episodes, and the maintenance of full remission without subsyndromal symptoms. Subsyndromal
symptoms and chronicity contribute significantly to long-term disability of individual patients (Coryell et al. 1993; Angst and Preisig
1995a, b) and thus also have a strong socioeconomic impact (Begley et al. 2001). Since it may be over-optimistic to expect any single
mood stabilising agent to be effective enough on all symptoms of bipolar disorder, an evidence based and skilful combination of two
or more medicines may increasingly become the strategy of choice (Post et al. 1996; Frye et al. 2000). Unfortunately, controlled data on different
combination strategies are still very limited. Combination treatment rests on informed choices of medicines, the properties being usually established only as monotherapy. Thus,
in the following section we will mainly focus on the published evidence for individual medicines, as confirmed by controlled trials or largescale naturalistic studies. When data are available, consideration will be given to combination treatment strategies.
2. Methods
A detailed
description of the methodology
is provided in the previously published
guidelines on bipolar depression (Grunze et al.
2002a). In
brief, the WFSBP set
up a task force on the biological treatment of bipolar affective
disorders including 55 members and a board with a chairman, two
co-chairmen and a
secretary. A
first draft of these
guidelines was
compiled by the board and sent out to all 55 members to collect their
comments on, additions to and corrections of this first draft. A second
draft was
compiled
incorporating the members’
suggestions. In order to ensure that the guidelines were as up-to-date
as possible, a second MEDLINE query and a search in abstract volumes of
international key
conferences was conducted
to include the latest information on treatment strategies up to January
2004. In addition,
other recently published expert statements and
guidelines were searched for additional references, e.g. the Austrian
(Kasper et al. 2003),
German (Grunze et al. 2002b), US (Zarin et al.
2002), British (Goodwin 2003), Danish (Licht et
al.
2003) and Australian
and
version of this
WFSBP guideline tries to incorporate the input of all the members of
the task force, although areas of uncertainty
and different emphasis may remain.
To grade evidence for any specific
treatment recommendation a modified version of the PORT recommendations
(Lehman and Steinwachs 1998) was used in these
guidelines, similar to the previous parts on acute
treatment. Four different levels of evidence were
defined:
Level A: Good research-based
evidence. This means that evidence for efficacy has been supported by
at least three
methodologically sound trials including at least one
placebocontrolled trial and at least two comparison trials with another
standard treatment. In
these trials, criteria such as sufficient sample
size, duration of trial, randomised and concealed distribution to
either treatment and
doubleblind conditions should have been met.
Level B: Fair research-based
evidence. On
the basis of trials, this includes evidence
from at least two randomised, double-blind (RDB) controlled trials
which, however, fail to
fulfil all the criteria above (e.g., small sample size
or no placebo control) or from one RDB study and
at least one prospective, large-scale
naturalistic study.
Level C: One randomised,
double-blind (RDB) study with comparator and one prospective open-label
(POL) study, or two POL studies
with >10 participants.
Level D: Recommendation based
on
prospective case studies with a minimum of 10 patients,
or large scale retrospective chart analyses
and support by expert opinion.
Classical bipolar I disorder is a severe,
often psychotic and usually intermittent
disorder. Most existing data relates to Bipolar I
patients. However in recent years the better
recognition
of less severe
elated states has expanded
our definition of bipolar disorder to include a
broad range of milder but still clinically
significant cases. This bipolar spectrum also
represents a challenge for maintenance treatment. The following
different longitudinal courses of
the illness will be considered here:
- bipolar
I disorder without rapid cycling
- bipolar
I disorder with rapid cycling
- bipolar
II disorder without rapid cycling
- bipolar
II disorder with rapid cycling
- schizoaffective disorder, bipolar type
These categories are not without
controversy. However, despite
belonging to the same spectrum, the longitudinal course of
Bipolar I and II disorder is distinct enough to allow
separation as
separate entities (Judd et
al. 2003), and while it is becoming apparent that to
define rapid cycling in a separate category is to
some degree artificial (Kupka et al. 2003,
2004), it is still consistently applied in prophylactic treatment
trials. It was also decided to
include schizoaffective disorder in these
guidelines because it may have more in common with
bipolar disorder
than with schizophrenia as
far as maintenance treatment is concerned (Marneros 2001). These
over-simple
categorisations, however, cannot do justice to the clinical
complexity of
treating bipolar disorder.
When finally choosing the first choice mood
stabiliser or other maintenance medication,
characteristics such as the following should also be
considered:
- Predominance of
mania or hypomanic episodes versus depressive episodes over
the course of illness
- Severity of episodes
- Whether previous episodes were or were not related to concurrent treatment with antidepressants or use or misuse of psychostimulants
- Long-term tolerability of the medication considered
- Patient
preferences
The term mood stabiliser needs some
clarification. No single agent shows equally good efficacy for all mood
disturbances
throughout the bipolar spectrum. Following the
suggestions
of Ketter and
Calabrese (2002), we have
here included, as mood stabilisers, medicines
that preferentially act on and prevent relapse
to only one pole of the illness (e.g., mania or
depression, without
ill effect on the
other). A note of caution has to be entered
concerning the duration of clinical trials. Studies
following strict scientific criteria do not usually
last longer than 12-18 months. Thus it is to a certain
degree problematic to claim prophylactic rather
than subacute maintenance properties for agents such as olanzapine or
lamotrigine.
A further limitation
of these international guidelines is the fact that, because not
all medicines are licensed and marketed in
every country, the reader has to take this into
consideration in
personal clinical
practice.
3. Bipolar I disorder without rapid cycling
Recommendations for
bipolar I disorder
without rapid cycling are the most evidence-based because, until
recently, maintenance
studies had been conducted only in this type of
illness.
Lithium
The prophylactic
efficacy of lithium became evident as early as the 1960s and early
1970s (Baastrup and Schou 1967; Prien et al.
1973a); for recent reviews, see Goodwin (2002) and
Müller-Oerlinghausen
et al. (2002).
Unfortunately, the prophylactic effect of lithium was probably over
stated, especially in trials
where responsive patients were re-randomised from
lithium to placebo,
due to discontinuation effects of lithium. A meta-analysis
applying the Cochrane methods, and controlling for
withdrawal studies, identified a total of nine studies
in unipolar and
bipolar affective disorder (Burgess et al. 2001): pooled analysis
indicated that lithium is nevertheless effective in preventing new
episodes in mood disorder. Although some aspects of the methodology of
the early trials can be questioned, their
positive findings have been independently supported recently by two
positive large-scale
placebo
controlled
double-blind randomised trials where lithium was used as a comparator
to lamotrigine (Bowden et al. 2003; Calabrese
et al. 2003a). Another recent large-scale
controlled
study (Bowden et al.
2000a), in which
lithium served as a comparator to valproate and placebo, failed to show
efficacy of lithium
in the primary outcome criterion (time to any mood
episode) due to a
combination of
methodological shortcomings and inadequate power. The secondary
analysis demonstrated lithium to
be effective for mania prevention. A further
metaanalysis of bipolar only trials, including five trials totalling
770 patients, indicated
that lithium is effective in preventing new
episodes in bipolar disorder, being preferentially
more
effective against
new manic episodes
(Geddes et al. 2004). Regarding optimal maintenance
blood levels, Gelenberg et al. (1989) found that
relapse rates were higher among patients with
plasma levels maintained at 0.4-0.6 mEq/L than
among patients with plasma levels maintained at
0.8- 1.0 mEq/L. However, tolerability problems
may limit the use of higher lithium doses for
longterm treatment. Together with other reasonably well-designed trials
against comparators
(e.g., carbamazepine
(Greil et al. 1997b, 1998; Hartong
et al. 2003) or imipramine (Prien et al.1973b) the
evidence for prophylactic
efficacy in bipolar I disorder for lithium
can be clearly graded as Level A. It is
probably more effective against manic than
depressive
relapse, so
where the illness
course is primarily driven by manic episodes
(more likely when manic episodes show
classical uncomplicated euphoria) it may be
preferred.
In clinical use,
however, the effectiveness
of prophylactic lithium across the clinical spectrum appears to be less
than suggested
by these controlled trials (Dickson and
Kendell
1986; Harrow et al.
1990; Goldberg et al.
1996; Keck and McElroy 1996; Maj et al. 1991,
1998). This efficacy-effectiveness gap may exist
not only for lithium,
but for all agents used
in
maintenance
treatment. Cohorts in clinical studies may not be fully representative
of
the clinical population with bipolar disorder
(Licht et al. 1997), since co-morbidity (Suppes et
al.
2001) or poor
adherence (Johnson and McFarland 1996; Scott 2002) complicate
treatment. Moreover, the structure of
clinical trials probably improves patient care and outcomes generally.
Therefore, putative predictors of favourable response to
lithium (family history of bipolar disorder, Mania-Depression-Free
interval course, no rapid cycling, no alcohol or drug abuse and,
especially, good adherence) should also be considered when recommending
treatment with lithium (Abou-Saleh and Coppen 1986; Maj 1992). In the
case of partial responsiveness to lithium, adding low doses (400 µg) of
folic acid has been suggested to improve the
outcome (Coppen et al. 1986). However, this
suggestion remained a point of controversy within the
task force, as its scientific rigour is low. Especially for lithium, an
increased
relapse risk after its sudden discontinuation has been described
(Mander and Loudon 1988) and
reinstituting lithium may not always be effective (Post et al. 1992;
Goodwin 1994). If
necessary, it is strongly recommended that lithium maintenance is
always tailed off slowly over some weeks or even months (Suppes et al.
1993). Informed adherence to treatment plans is clearly necessary in
bipolar patients, and
formal psychoeducation can improve outcomes (Colom et al. 2003).
Attention to patient
information and adherence should be mandatory for patients selected for
long-term
prophylactic treatment.
Antiepileptic medications
As already
mentioned, several antiepileptic medicines have also been studied for
maintenance efficacy in classical bipolar I disorder.
Lamotrigine
Recent guidelines
have recommended lamotrigine for the acute treatment of
bipolar depression
(Zarin et al. 2002; Calabrese et al. 2004), despite the absence of
unambiguous
proof in controlled
studies (Calabrese et
al 1999b). In contrast, two recent studies
(Bowden et al. 2003; Calabrese et al. 2003a) have demonstrated the
prophylactic efficacy of
lamotrigine. Both
studies (in patients
recovering from mania or from depression respectively) were conducted
under double-blind conditions using a three-arm design with placebo and
lithium as
comparators. The patient sample
was moderately enriched for likely lamotrigine response and
tolerability. They were
stabilised during an open run-in phase with
lamotrigine
in addition to any
other indicated
psychotropic agents. Patients were required to maintain stability on
lamotrigine alone after
withdrawal of other acute treatments. Thus, patients
experiencing
side-effects with lamotrigine
or who were unresponsive to acute treatment
with lamotrigine may not have been randomised. The main result of both
studies was that
lamotrigine was
statistically superior to
placebo (as was lithium) for the primary outcome – “time to
intervention for a mood episode”. Secondary analysis revealed that this
overall
outcome was mostly
due to a preventative
effect of lamotrigine against depressive relapse independent of whether
the index episode
was manic or depressive. Lamotrigine was less
effective in
preventing new manic episodes,
but in a combined analysis it was still
significantly better than placebo (Goodwin et al. 2004).
In contrast, lithium was more effective in
preventing new manic
episodes, but less effective in preventing bipolar depression.
These apparently complementary properties may support the combination
of lithium with
lamotrigine in
clinical practice although,
so far, there are no controlled studies supporting
this approach. In addition to these controlled trials
there exist data from open, successful long-term lamotrigine treatment
(Walden et al. 1996, 2000;
Calabrese et al. 1998; Suppes et al. 1999a; Bowden
et al. 1999). In summary, we
would grade lamotrigine as Level A
concerning prophylactic efficacy, with its
main strength being the prevention of new
depressive episodes (Goodwin et al.
2004).
Valproate
Expectations for
valproate as prophylactic medication were quite high because several
open, but in part, randomised studies
(Lambert 1984; Denicoff et al. 1997a; Hirschfeld et
al.
1999; Emrich and
Wolf 1992; Hayes 1989; Vencovsky
et al. 1984; Solomon et al 1998) appeared positive. However, the only
randomised double-blind placebo-controlled study
failed to
demonstrate a statistical
benefit. The prophylactic efficacy of neither valproate,
nor lithium, which was used as an active
comparator, could be shown for the primary outcome
parameter (“time to
any mood episode”) (Bowden et al. 2000a). Secondary analysis
(time to a specific mood episode) suggested
efficacy of valproate, especially in preventing a new
depressive episode.
This finding is also
discussed in a recent Cochrane analysis (Macritchie
et al. 2001). Extension of an acute mania trial
(Tohen et al. 2003) over 47 weeks permitted
comparison of two small parallel groups who had
remitted on valproate or olanzapine (Tohen et al.
2002a). Double-blind conditions were maintained throughout the study.
At endpoint, there
was no significant difference between groups in
the number of manic relapses (olanzapine 41 %, valproate 50 %), or in
the median time to relapse. There were, however, differences
in secondary outcomes such as affective
stability and measures of cognitive function in
favour of olanzapine.
Thus, together
with the evidence from
open long-term studies we would grade
the level of evidence for prophylactic efficacy
for valproate as B. It may be especially
helpful for
preventing new
depressive episodes,
although further evidence on this point is needed.
Carbamazepine
Carbamazepine has
been widely used as a
firstline alternative to lithium maintenance
treatment in patients not responding sufficiently to or not tolerating
lithium, especially in
and Japan. There is
some evidence that carbamazepine is
superior to placebo in controlled trials (Okuma et al. 1981) and
it has also been claimed that its usefulness may
be comparable
to lithium (Placidi et al. 1986; Lusznat
et al. 1988; Denicoff et al. 1997b). However, these studies employed
methodology that would not now be supported, and very small numbers of
patients, so the results
are difficult
to interpret (Dardennes et al. 1995). There have been two more recent
impressive head-to-head comparisons of carbamazepine with lithium. The
key finding of the
MAP-Study was that for patients with a history of
classical euphoric mania, lithium was superior to carbamazepine
(Greil et al. 1997b). This
was confirmed in a recently published study of patients receiving
maintenance treatment
for the first time. Hartong et al. (2003)
compared lithium and carbamazepine under doubleblind randomised
conditions for two years. At the endpoint, lithium was outperforming
carbamazepine on dropout rates due to
relapse. In fact, there was a linear decline of the
efficacy of carbamazepine, whereas lithium patients tended to either
relapse early
(carbamazepine appeared superior for the first six months)
or remain stable for the duration of the
study. The lower efficacy of carbamazepine in the
second year cannot be explained by autoinduction
of the metabolic pathway of carbamazepine, because dosages of both
lithium and carbamazepine were based on plasma levels throughout the
study. A significant number of the patients not responding to lithium
may have atypical features – for example, dysphoria rather
than euphoria, bipolar II or ‘not otherwise
specified’diagnoses, mood incongruent delusions and
comorbidities. In a secondary analysis of the MAP data, this atypical
group showed a better outcome with carbamazepine compared to lithium
(Greil et al. 1998). Thus, on a formal
level, carbamazepine would fulfil the study
requirements to be graded as Level B. Carbamazepine may
be more
useful in atypical
manifestations but will
usually show poorer long-term efficacy than
lithium. As a significant disadvantage, it can also
induce the metabolism of antidepressants,
antipsychotics and other antiepileptic agents used as mood stabilisers,
thus leading to a loss of
efficacy of co-medications (Spina et al. 1996;
Hesslinger et al. 1999). Thus, its use is impractical in
certain combination treatment regimens.
Atypical antipsychotics
Controlled evidence
that antipsychotics
show long-term efficacy is currently available
only for olanzapine (Kasper et al. 2002). Olanzapine showed
prophylactic efficacy compared to
placebo in a
12-month study in patients who remitted from mania under olanzapine
treatment. Although both the prevention of new manic and depressed
episodes was
significantly
better for
olanzapine than placebo, the
effect was more pronounced for the prevention of manic relapses (Tohen
et al. 2004). In a
large, one-year randomised double-blind comparator
study versus
lithium, olanzapine showed at
least comparable efficacy to lithium - perhaps
even superiority in preventing a manic relapse (Tohen et al. 2002c).
Remembering the
valproate
maintenance study (Bowden et al. 2000a), the absence of a placebo arm
limits
the interpretation of this trial. Olanzapine as
an add on to lithium or valproate also performed
significantly better
in relapse prevention compared to placebo in a double-blind, 18-month
extension of the respective acute
mania study (Tohen et al. 2002b). Thus, olanzapine
may be
rated as having Level A
evidence for prophylactic treatment with a
probably greater efficacy against the manic
pole of the illness. Its profile is thus
similar to lithium. The
only qualification
must be that the key maintenance studies were targeting a
population of patients acutely responsive to olanzapine, a selection
bias even more
pronounced
than in the
lamotrigine studies. Accordingly, the relative prophylactic
efficacy of olanzapine in a non-selected sample remains unclear.
Clozapine has been
in use as an off-label
agent in refractory bipolar disorder for more
than 30 years. In this indication it has shown
efficacy mostly in small, investigator-initiated
trials for both acute and maintenance treatment
(Suppes et al. 1999b). In addition, several case
reports (Puri
et al. 1995; Hummel et al. 2002; Kurz et al. 1998;
Zarate et al. 1995b; Banov et al. 1994) support long-term efficacy of
clozapine.
Despite its widespread clinical use, the
published evidence for clozapine for
prophylactic treatment of classical bipolar I
disorder is
limited
(Level C). Risperidone, at least as an add-on
medication, may be useful in bipolar disorder
for long-term use, despite the absence
of randomised controlled trials
(Level D). It has been added
to mood stabilisers in
symptomatic patients under open conditions (Vieta et
al. 2001). This regimen not only improved manic and depressive symptoms
significantly, but
also appeared to reduce recurrence for the next
six months.
Antidepressants
In bipolar patients
where recurrent
depressive episodes dominate the clinical course, the
longterm use of antidepressants in combination with a mood stabiliser
may be considered.
On
the other hand,
monotherapy, especially
with a tricyclic antidepressant, should be
discouraged. In the Prien study (Prien et al. 1973b,
1984) comparing lithium and imipramine,
imipramine was effective in preventing new depressed episodes but was
associated with a high
switch risk into mania. The situation may be
different with newer antidepressants such as SSRIs,
or combined norepinephrine and serotonin
reuptake inhibitors, e.g. venlafaxine and milnacipran, and bupropion.
Naturalistic
data from the
Further treatment options
Other treatment
options, e.g. calcium
antagonists, are usually only considered for patients with a rapid
cycling course of bipolar
disorder (Post et al. 1990). Despite the high risk
of
persistent
tardive dyskinesia (Mukherjee et al. 1986), long-term use of typical
neuroleptics, especially as a depot formulation, is
sometimes considered in non-compliant bipolar I
patients, but is mostly reserved for patients with
pronounced psychotic symptoms or schizoaffective disorder. Littlejohn
et al. (1994) compared
the duration of affective episodes in periods
on or off depot antipsychotics for 18 bipolar
patients who had been treated for a mean of 8.2
years. Patients in this naturalistic study
suffered fewer relapses and spent significantly less time
in hospital (P = 0.001) for treatment of
manic, depressive or mixed affective illness
during treatment with depot antipsychotics.
However, a recent double-blind, placebo-controlled
trial showed significantly greater rates of
emerging depression when a typical neuroleptic was maintained after
remission from mania
(Zarate and Tohen 2004).
Maintenance ECT is
believed to be effective
on the limited evidence of single case reports
and small case series in bipolar disorder
(Kramer 1999; Gupta et al. 1998; Rabheru and Persad
1997; Vanelle et al.
1994; Shapiro et al.
1989), but it is also associated with a moderate
switch risk when given without concomitant mood stabiliser treatment
(Angst 1985).
Patient
psychoeducation through group, individual or family approaches can play
a crucial part in the long-term success of maintenance treatment with
medicines.
Increased
adherence, improved
awareness of relapse prodromes and effective action
planning are central components of such
interventions. A detailed review of psychological
interventions is
beyond the scope of
this review, but the
key issue is how the lessons from quite
specialised interventions can be generalised to good clinicalcare.
4. Classical bipolar I disorder with
rapid cycling
Lithium
Dunner and Fieve
(1974) found that patients with four or more episodes per year were on
the whole less responsive to lithium than those with fewer episodes,
and coined the term
“rapid cycler” for this subgroup. In non-selected populations rapid
cyclers may constitute
15-20% of all bipolar patients; however, in specialised clinics they
may make up more
than
50% (Suppes et al.
2001). Thus, rapid
cycling especially constitutes a particular
challenge for expert settings rather than for general
practitioners. After Dunner and
Fieve’s study, opinion hardened that lithium is not useful in
rapid cycling bipolar patients. However, this may have been premature,
and rapid cycling
patients may be less responsive to most treatments.
In fact, in Dunner and Fieve’s study, lithium appeared to prevent new
manic episodes in rapid cycling patients but it had no effect
on new depressive episodes. In other small
open trials in rapid cycling patients (e.g.,
Walden et al. 2000), a small prophylactic effect of
lithium was again observed (although it was
inferior to the active comparator, lamotrigine). Thus, we
may grade the efficacy of lithium
for the prevention of new manic episodes
in rapid cycling bipolar disorder as C, and
it may not be efficacious in preventing new
depressions.
Valproate
Evidence for the
efficacy of valproate in
bipolar I rapid cycling comes purely from open
studies. The largest study by Calabrese et al.
(1993) showed prophylactic efficacy in rapid
cycling patients against new episodes of typical
mania, mixed states and to some degree also
bipolar depression. This impression is also
supported by other smaller open studies (Jacobsen 1993; Sharma et al.
1993; Walden and Grunze
2002). The only randomised, double-blind study so
far, comparing valproate and lithium in bipolar
I and II rapid cycling, failed due to a lack
of power (Calabrese et al. 2003b): only one in four patients was
successfully stabilised at
baseline and thus eligible to enter the randomised
study phase. Numerically, valproate was superior
to lithium. Thus we may
conclude evidence for valproate is, at best, Level C in
the maintenance treatment of bipolar I rapid
cycling.
Lamotrigine
The most reliable
controlled data in rapid cycling are available for lamotrigine. In a
large randomised, placebo-controlled six-month study lamotrigine was
significantly
superior to
placebo for survival
in study. However, it
was not positive for the primary efficacy
measure (time to intervention). Secondary analysis
also showed the effect to be confined to bipolar
II patients and not bipolar I (see related
section). Another as yet unpublished controlled trial failed to show a
significant improvement
with lamotrigine compared to placebo. Open data
appear to support a
beneficial effect of lamotrigine in preventing bipolar I rapid
cycling (Kusumakar and Yatham 1997; Fatemi et al. 1997; Bowden et al.
1999; Walden et al.
2000).
We may therefore
conclude that lamotrigine
monotherapy has Level D evidence
for treating bipolar I rapid cycling.
Carbamazepine
The evidence for a
prophylactic efficacy of carbamazepine in rapid cycling is
contradictory. The only controlled study comparing lithium and
carbamazepine failed to show a better
effect of carbamazepine (Okuma 1993). Some open trials gave positive
hints (Di Costanzo and Schifano 1991; Strömgren and Boller 1985), while
others did either not or only
partially
replicate this
finding (Joyce 1988; Post et
al. 1990). None of these studies was performed under controlled
conditions so that the use of
carbamazepine in bipolar I rapid
cycling
relies on
case studies and expert
opinion (Level D).
Calcium antagonists
Two studies
(Pazzaglia et al. 1993, 1998) conducted under double-blind conditions
with an on-off-on design supplied evidence for a prophylactic efficacy
of the calcium
antagonist
nimodipine, even in
previously refractory bipolar I rapid cycling patients (Level
B). The clinical
usefulness may be limited by the
short half-life - a daily dosage of more than 240
milligrams is
usually needed, corresponding
to eight tablets given in three doses.
Nevertheless, this option should be considered in
patients where other treatment attempts have failed.
Atypical antipsychotics
As already
mentioned, there is some
evidence for prophylactic efficacy of clozapine in
rapid cycling
patients (Hummel et al. 2002; Frye et al. 1996;
Suppes et al. 1994; Calabrese et al. 1991). Again, these studies are
rather small and questionable from a methodological point of view.
However, given that these patients
are difficult to treat, the option
of clozapine in
refractory
rapid cycling patients
should always be considered, although the
grading according to the level of evidence
is currently D. For other atypical
antipsychotics, data are
still not adequate
to make any judgement concerning prophylactic efficacy.
Combination treatment
Especially in rapid
cycling, combination of mood stabilisers is the clinical rule, not
the exception. This reflects the generally
disappointing response to monotherapies. Most data
on combination
treatments for prophylaxis
of bipolar disorders come from open studies,
but blinded, prospective studies have also been conducted
(Denicoff et al. 1997b; Solomon et al. 1997). In the landmark, albeit
underpowered, study by Denicoff et al. the prophylactic
efficacy of lithium, carbamazepine and their combination was compared.
Fifty-two outpatients
who met DSM-III-R
criteria for bipolar
illness were randomly assigned in a double-blind design either to start
with lithium or carbamazepine, followed by a crossover to
the
opposite medication
and to the combination with each treatment lasting one year. The
percentage of the evaluable patients who
had marked or moderate improvement on the
Clinical Global
Impression scale was 33.3%
on lithium, 31.4% on carbamazepine and 55.2%
on the combination treatment (not significantly different). Patients
with a past history of
rapid cycling did poorly on monotherapy (28.0% responded to lithium;
19.0% responded to carbamazepine), but significantly better on
the combination (56.3%, p < .05). A small
study by Solomon in 12 non-rapid cycling bipolar I patients tested the
efficacy of lithium
plus valproate versus the efficacy of lithium
alone in continuation and maintenance treatment over
one year (Solomon et
al. 1997). Patients
treated with the combination of lithium and valproate were
significantly less likely to suffer a
relapse or recurrence (p = .014), but were also
significantly more likely to suffer at least one moderate
or severe adverse side effect (p = .041).
Thus, clinicians prescribing combination
treatment must weigh any benefit against this
potential cost.
With the
studies available, the
evidence for combination treatment may be
graded as Level C. The choice of combination
will often depend on a clinical judgement as
to the
balance of
morbidity between the
poles of the illness.
5. Bipolar II disorder without rapid cycling
Bipolar II disorder
has just recently
become a focus of research interest, especially with
the emergence of new agents such as
lamotrigine. This interest is also due to increasing
awareness of the high prevalence of bipolar II and, especially,
spectrum disorders (Akiskal et
al 2000; Angst and Gamma 2002). Unfortunately, this new interest in
bipolar II disorder is
not so far reflected in large randomised and
placebocontrolled studies (Suppes et al. 2002). Current treatment
guidelines for bipolar II
disorder, including the American Psychiatric
Association guidelines (American Psychiatric
Association 2002) and the
bipolar II patients
with a rapid cycling
course (see relevant section). As long as clear evidence is lacking
that a specific mood stabiliser is superior to
others in bipolar II disorder without rapid cycling,
a general rule may be to continue the mood stabiliser that was
effective in acute
treatment. For those patients receiving a mood
stabiliser for the first time, general illness features
may also guide the selection of a treatment. Bipolar
II disorder is characterised not only by more
interepisodic (Benazzi 2001) and episodic morbidity, and mostly severe
depressive episodes
(Vieta et al. 1997b; Tondo et al. 1998), but also by
a high degree of suicidality (Vieta et al. 1997a;
Rihmer and Pestality 1999). Thus, those
moodstabilisers which show preventive effects mainly on depression in
bipolar I disorder, such
as lamotrigine, or suicidality, such as
lithium (Tondo and Baldessarini 2000), may be the agents of choice. In
addition, lithium
seems to be of equal efficacy in uncomplicated
bipolar II patients compared to bipolar I disorder. The use of
tricyclics in maintenance
treatment is discouraged due to their probable
propensity to induce rapid cycling (Altshuler et al.
1995), however, long-term treatment with modern antidepressants should
not be ruled out, although the evidence is still preliminary (Altshuler
et al. 2003). Switch rates into
mania for bipolar II patients treated with modern antidepressants are
still a subject of
controversy (Ghaemi et al. 2001), but if switches occur
they are more likely of moderate severity
(Benazzi1997).
Currently,
only lithium reaches a
Level C classification, and the evidence
for carbamazepine is no better than Level D
criteria.
6. Bipolar II disorders with rapid cycling
Secondary analysis
of a placebo-controlled, double-blind, randomised study with
lamotrigine in rapid cycling patients suggested that it may be of
special value in bipolar II
patients
(Calabrese et al.
2002). For the 52
patients with bipolar II disorder, the median time to
intervention was significantly greater in those
receiving lamotrigine (17 weeks vs. seven weeks).
However, the positive effect of lamotrigine in
bipolar II disorder was a post hoc finding
and related to reduction of depression only. Together with supporting
evidence from open case series (Calabrese et al. 1999; Suppes et al.
1999a; Fatemi et al. 1997) a Level C
grading may be appropriate for lamotrigine.
For valproate, some
evidence of usefulness comes from the trial by Calabrese and
Delucchi (1990) already cited. Thirty of the 55
individuals with rapid cycling had a bipolar II
disorder.
Interestingly, responsiveness
to valproate appeared to be slightly better in
those bipolar II patients compared to bipolar I
(Level D)
The positioning of
schizoaffective
disorder,bipolar type, between schizophrenia and affective disorders
has always been a
matter of controversy and threatens to shift with
every
new diagnostic
manual. Furthermore, the diagnosis of schizoaffective disorder, even
in good studies, is not reliable: e.g., the
inter-rater reliability in Maj’s study (2000) for
schizoaffective disorder had a kappa of 0.22. It is also of some
concern that schizoaffective
treatment data are almost exclusively provided as sub-group analyses
from schizophrenia
trials.However, in order not to overlook this patient group, we will
briefly summarise
the evidence.
Referring to trials
primarily addressing
relapse in schizophrenia, from atypical
antipsychotics, namely, risperidone, olanzapine,
ziprasidone and aripiprazole, schizoaffective patients
showed prophylactic
efficacy to a similar
degree.However, the primary outcome was almost exclusively focused on
emerging psychotic,
not mood symptoms, which makes the findings
hard to interpret
fully. Given the acute antimanic efficacy of all atypical
antipsychotics, however, it appears likely that positive
effects on mood symptoms can also be observed in
schizoaffective
patients. The efficacy of both lithium (Tress and
Haag 1979) and carbamazepine (Vovin et al. 1984) has been investigated
in open trials
showing efficacy for both agents. In an open, randomised, multicentre
study (the MAP
study), the prophylactic efficacy of lithium and carbamazepine was also
compared (Greil et
al. 1997a). A total of 90 ICD-9 schizoaffective patients were included
in the maintenance phase (2.5 years). They were diagnosed according
both to Research Diagnostic
Criteria (RDC) and DSM-III-R and classified into subgroups. Outcome
criteria were hospitalisation, recurrence, concomitant psychotropic
medication and adverse effects leading to discontinuation. There were
more noncompleters under carbamazepine than under lithium (p = 0.02).
Survival analyses
demonstrated no significant differences between lithium and
carbamazepine in treatment
outcome. Patients’
ratings of side effects
(p = 0.003) and treatment satisfaction (p =
0.02) favoured carbamazepine. Following the RDC criteria, patients of
the schizodepressive
and
non-classifiable
type did better with carbamazepine than lithium (p = 0.055 for
recurrence), whereas in the schizomanic patients, they were equipotent.
Applying
DSMIII-R, carbamazepine demonstrated superiority in the patient group
with more
schizophrenialike or depressive disorders (p = 0.040 for recurrence),
but not in patients fulfilling
the
DSM-III-R criteria
for schizoaffective
disorder, bipolar type. This is the best evidence
currently available for prophylactic efficacy of mood stabilisers in
schizoaffective disorders.
Lithium and carbamazepine seem to be equal
alternatives in the maintenance treatment of broadly defined
schizoaffective disorders. However,
in subgroups with depressive or
schizophrenia-like features, and in long-term tolerability,
carbamazepine may be superior. These findings are in line with previous
reports that lithium may
be more effective in schizoaffective disorder, mood-dominant type
(Müller-Oerlinghausen et al. 1989), than in schizophrenia-dominant
patients (Lenz et al. 1989). In patients not responding sufficiently
with
lithium prophylaxis,
augmentation with
carbamazepine may be successful (Bocchetta et al.1997).
Both
carbamazepine and lithium in
these indications may be ranked as Level
C. Valproate has also demonstrated
prophylactic usefulness in schizoaffective
disorder, but only on the basis of case series
and open trials (Puzynski and Klosiewicz 1984;
McElroy and Keck 1993; Keck et al. 1996)
(Level D).
Despite the positive
findings with mood stabilisers, concomitant treatment with an
antipsychotic as well often appears prudent
in schizoaffective disorder. Most experience
derives
from open studies
where atypical
antipsychotics appear to be a more promising treatment
than typical neuroleptics (Ghaemi and Goodwin 1999). Clozapine has
traditionally been
used in the prophylaxis of schizoaffective disorder (Suppes et al.
1999b; Ciapparelli et al.
2000; Hummel et al. 2002) (for a review of the
earlier literature, see also Zarate et al. 1995a).
A
subgroup analysis of
a large controlled
Phase III trial revealed that olanzapine is more
efficacious than haloperidol in relapse prevention of schizoaffective
disorder (Tran et al. 1999).
Similarly,
risperidone appears more
effective and tolerable than haloperidol as
demonstrated by a double-blind, large-scale prospective
maintenance trial for schizophrenia and schizoaffective disorder
(Csernansky et al. 2002). However, these two controlled trials lasted
only one year, and more clinical, long-term experience is needed for a
definite
judgement.
For example, a
recent report on quetiapine treatment raised the question of loss of
efficacy and rebound psychosis over three years (Margolese et al.
2002). When combining a mood stabiliser and
antipsychotics, lithium or valproate may be the primary choice as the
mood stabiliser. The induction of cytochrome P450-dependent metabolism
by carbamazepine often leads to insufficient serum levels of
antipsychotics
(Spina et
al. 1996; Hesslinger et al. 1999; Yatham 2000). In addition to the
range of medicines
discussed above, maintenance ECT may also be an effective treatment
(Swoboda et al. 2001).
8. Additional
psychopharmacological treatment options
Besides the cited
pharmacological treatment options, there is a wide range of
additional experimental treatments or nutritional
agents currently under investigation for mood
stabilising
properties. This
includes, e.g., thyroid supplementation in patients with
rapid-cycling bipolar disorder (Bauer and Whybrow 1990), novel
anticonvulsants like oxcarbazepine
(Grunze and Walden
2002) or nutritional supplements, e.g. omega-3 fatty acids
(Severus et al. 1999). For reasons of space and the
currently limited quality of available study data, we
will not discuss these options further. We note
that they may, nevertheless, prove important in
the future.
9. When is maintenance
treatment indicated?
There is no
controlled prospective study to indicate when maintenance treatment
should start. Several retrospective chart analyses
suggest that with every episode the length of the
subsequent
symptom-free interval decreases (Kessing 1998; Roy-Byrne et al. 1985;
Angst 1981; Zis et al. 1980). For lithium, there
is also some recent evidence that prophylactic
efficacy may decrease with a longer delay between
onset of illness and initiation of treatment
(Franchini et al. 1999; Garcia-Lopez et al. 2001).
However, there are also contradictory data
(Baldessarini et al. 1999b) and it remains unclear whether
this phenomenon is only true for lithium or also
for other treatment modalities. These findings justify starting
maintenance treatment as
soon as possible after the diagnosis has been established. However, not
all patients
would suffer from an additional episode (Goodwin 2002), and the
acceptance of long-term treatment by many patients is low at this
early stage so to prescribe lithium at all risks
its premature discontinuation. Sudden
discontinuation, especially of lithium, may harm
patients more than
having never been on prophylactic treatment (Baldessarini et al. 1999c;
Goodwin 1994) and increase suicide
risk (Baldessarini et al. 1999a). Whereas many European guidelines
usually recommend waiting for at least a second
episode of illness, and only recommend maintenance treatment if these
episodes occur within a
rather short time interval (e.g. three years), US guidelines favour
commencement of
maintenance treatment with the first manic episode (Sachs
et al. 2000; Bowden et al. 2000b). Compromising between these
recommendations, the Dutch guidelines consider the number of episodes
and variables such as severity
and positive family history of bipolar disorder suggesting an increased
genetic risk (Nolen
et al. 2001). Thus, if the first episode is manic,
of disruptive severity, and there is a family
history, they recommend to consider seriously the
start of maintenance treatment. Otherwise, with
two episodes (one of them manic), maintenance treatment should be
initiated if at least
one is of particular severity or the patient has a
positive family history. With the third episode, prophylaxis should
always be considered
(Figure 1). Whatever the advice from doctors may
be, the limiting consideration at this stage is
often the attitude of patient and family. Concerning the duration of
prophylactic
treatment, there is a consensus among experts that it should be
continued life long whenever possible, although there are no
discontinuation studies targeting this question.
Both the broadening
of the bipolar spectrum and the increasing awareness of the limitations
of lithium (Maj et al. 1998), and the rest
of our current treatment portfolio (Frye et al.
2000),
especially with
respect to bipolar
depression (Akiskal et al. 2000), gave rise to the
development of new treatment strategies. Some
antiepilectic medications and atypical antipsychotics
may be helpful
alternatives, either in monotherapy or combination treatment. In other
than classical manifestations, they
may even be the first choice treatment.
However, more controlled data, especially beyond two years (similar to
what is available for
lithium and carbamazepine) and long-term clinical experience are
needed, and several
important questions, e.g. anti-suicidal properties,
still have to be addressed (Goodwin et al. 2003).
Table 1 summarises current, but only to some extent evidence based,
recommendations for
maintenance
treatment of the
different forms of bipolar disorder. It depicts a suggested
sequence of use of medications. With a severe and refractory course of
bipolar disorder,
combination treatment should be initiated early in some instances, e.g.
often in schizoaffective
disorder and rapid cycling patients, right from the beginning. We can
reasonably claim to have identified rational approaches to long-term
treatment
in bipolar patients, based loosely on
evidence, albeit of varying strengths. The future
development of combination treatments will only be optimised if we move
beyond mere reason to high quality evidence from large scale
trials in representative patient samples.
Quand un
traitement de maintenance est-il nécessaire ?
