There is no doubt that antipsychotic treatment is valuable in acute psychotic episodes. There is also considerable evidence that antipsychotics are useful for those with continuing positive psychotic symptoms. Indeed, the introduction of antipsychotics in the 1950s is credited with having contributed to the return to the community of many previously chronically ill patients from asylums, and the running down and eventual abolition of these institutions. Reference Shorter1
However, a new era was ushered in by Leff & Wing in 1971 when they reported an randomised controlled trial (RCT) of the prophylactic value of antipsychotics in preventing the relapse of people with schizophrenia. Reference Leff and Wing2 Subsequently, many other RCTs confirmed that relapse was less common in those who continued treatment with antipsychotics. Although most of these trials lasted for less than 1 year, Reference Leucht, Tardy, Komossa, Heres, Kissling and Salanti3 it became common practice for those who had received a diagnosis of schizophrenia to remain on antipsychotics for many years.
However, increasing doubts have been raised about the long-term prophylactic use of antipsychotics. Reference Moncrieff4 As with any drug treatment, the beneficial effects have to be balanced against the potential side-effects. This paper will therefore address five troublesome issues: (a) effects of antipsychotics on physical health, (b) effects of antipsychotics on brain structure, (c) efficacy of long-term antipsychotic use, (d) antipsychotic-induced dopamine receptor supersensitivity, and (e) treatment-resistant schizophrenia (TRS).
People with schizophrenia die earlier than the rest of the population. Reference Walker, McGee and Druss5,Reference Brown, Kim, Mitchell and Inskip6 Studies from several countries show that the trend of mortality rates in people with psychosis has not followed the same downward course as that of the general population, resulting in a widening disparity. Reference Hoang, Stewart and Goldacre7–Reference Laursen, Nordentoft and Mortensen10
Suicide accounts for part of the excess mortality, although the most obvious remediable cause is the excessive use of tobacco; illicit drug use, poor diet and lack of exercise are also implicated. However, the possibility that antipsychotics also contribute has attracted much attention recently. Antipsychotics can cause cardiac arrhythmias Reference Frommeyer and Eckardt11,Reference Ray, Chung, Murray, Hall and Stein12 and induce dyslipidaemia and obesity, with activation of the cascade towards insulin resistance and diabetes. Reference Henderson, Vincenzi, Andrea, Ulloa and Copeland13–Reference De Hert, Detraux, van Winkel, Yu and Correll15 Most second-generation antipsychotics (SGAs) confer a higher cardiometabolic risk than first-generation antipsychotics (FGAs), and some authors point out that the switch to SGAs from the mid-1990s has paralleled the widening of the mortality gap. Reference Saha, Chant and McGrath8
However, there is still dispute over whether long-term antipsychotic prescription actually increases overall mortality. For example, data from Swedish nationwide registers show a lower mortality in patients with chronic schizophrenia receiving long-term antipsychotics than in those with similar duration of illness but no antipsychotic exposure. Reference Torniainen, Mittendorfer-Rutz, Tanskanen, Bjorkenstam, Suvisaari and Alexanderson16 It could be that antipsychotics reduce suicides or facilitate better self-care – or simply, as Mace et al Reference Mace, Dzahini, Cornelius, Anthony, Stewart and Taylor17 suggest, that taking antipsychotics ‘may be a surrogate marker for overall healthy behaviour’.
In spite of the above controversy, there is enough evidence concerning the adverse effects of antipsychotics on physical health to compel psychiatrists to act. There are three obvious ways to address the risk. First, to prescribe, wherever possible, antipsychotics with a lower propensity to induce weight gain. Second, to facilitate a healthy diet and lifestyle from the start of treatment. Third, to reduce the use, or at least the dosage, of long-term antipsychotics. In implementing this last strategy, one must, of course, consider the possible impact on the mental health of each patient.
Structural brain changes
Many people with schizophrenia show subtle developmental brain abnormalities at the onset of their illness. Reference Murray and Lewis18 However, further brain changes occur in the years following the onset of the illness. These were interpreted initially as reflecting an intrinsic degenerative schizophrenic process. Reference van Haren, Cahn, Hulshoff Pol and Kahn19 However, a number of reports have concluded that the majority, if not all, of these brain changes are a consequence of environmental factors; among these factors are antipsychotic medication, Reference Navari and Dazzan20–Reference Vita, De Peri, Deste and Sacchetti23 illicit drug use, Reference Rais, Cahn, van Haren, Schnack, Caspers and Hulshoff Pol24 cigarette smoking, Reference Karama, Ducharme, Corley, Chouinard-Decorte, Starr and Wardlaw25 obesity and diabetes. Reference Katon, Pedersen, Ribe, Fenger-Gron, Davydow and Waldorff26 Animal studies have confirmed the effects of antipsychotics in decreasing cortical volume and have shown, at least in monkeys and rats, that the changes are not due to a loss of neurons. Reference Dorph-Petersen, Pierri, Perel, Sun, Sampson and Lewis27–Reference Konopaske, Dorph-Petersen, Pierri, Wu, Sampson and Lewis29
Long-term use of high-dose FGAs does carry a risk of decreasing cortical volume and increasing ventricular volume; low-dose antipsychotics and SGAs may carry less risk. Reference Vita, De Peri, Deste, Barlati and Sacchetti30 These effects appear to be reversible on stopping antipsychotics. Reference Boonstra, van Haren, Schnack, Cahn, Burger and Boersma31,Reference Vernon, Natesan, Crum, Cooper, Modo and Williams32 Some preclinical behavioural studies have suggested that chronic administration of antipsychotics can impair memory-related task performance, as well as spatial and working memory. Reference AV, Gearhart, Warner, Zhang, Bartlett and Middlemore33,Reference Terry, Hill, Parikh, Evans, Waller and Mahadik34 Although the changes in grey matter observed in schizophrenia are correlated with deficits in cognitive domains, the latter have not been specifically linked to antipsychotics. Reference Gur, Turetsky, Bilker and Gur35
In short, there is still no clear link between the antipsychotic-associated changes seen with magnetic resonance imaging (MRI) and cognitive decline or functional impairment. Therefore, until more research evidence is available, the jury remains out on what, if anything, we should do about the effects of antipsychotics on brain structure.
Do antipsychotics improve long-term outcomes?
Many RCTs have compared outcomes in those patients with schizophrenia continuing antipsychotics with those in whom the drugs were stopped. These ‘discontinuation’ studies uniformly show a greater likelihood of relapse in the latter group. Reference Taylor, Cavanagh, Hodgson and Tiihonen36 Thus, a meta-analysis of 65 trials Reference Leucht, Tardy, Komossa, Heres, Kissling and Salanti3 concluded that ‘antipsychotic maintenance treatment substantially reduces relapse risk in all patients with schizophrenia for up to 2 years of follow-up’. However, few studies have lasted longer than this, leading Leucht et al to point out that ‘nothing is known about the very long-term effects of antipsychotic drugs compared to placebo’. Reference Leucht, Tardy, Komossa, Heres, Kissling and Davis37 Thus, long-term ‘maintenance’ treatment with antipsychotics is based on hope rather than evidence.
New studies have shed light on this issue. Harrow Reference Harrow, Jobe and Faull38 repeatedly assessed the course of 70 young patients diagnosed with schizophrenia. Those 15 not prescribed antipsychotics over the 20 years showed significantly fewer psychotic symptoms than those 25 continuously on antipsychotics. Moilanen and colleagues, Reference Moilanen, Haapea, Miettunen, Jaaskelainen, Veijola and Isohanni39 who followed up 74 patients with first-episode psychosis (FEP) in Finland, reported that after 10 years, the 24 not receiving medication had better clinical outcomes than those receiving antipsychotics. At the ten-year follow-up of 274 FEP patients in the UK ÆSOP sample, Reference Morgan, Lappin, Heslin, Donoghue, Lomas and Reininghaus40 Morgan (personal communication, 2016) found that 18% of those who received a diagnosis of schizophrenia had not taken antipsychotics for two years and had no psychotic symptoms. Similarly, Wils & Nordentoft (unpublished, further details available from M.N. on request) followed up 496 patients from the Danish OPUS study for 10 years after their first episode of schizophrenia spectrum disorder; 30% had remission of psychotic symptoms and were not taking antipsychotics.
There is a major confounder in all these studies, in that they were without randomisation. Consequently, those patients who were able to stop their medication may have had a milder illness than those who continued. Nevertheless, these studies demonstrate that a significant proportion of patients with psychosis manage to come off antipsychotic medication without detriment in the long term.
In a study less open to bias, Wunderink and colleagues selected a group of FEP patients who had been initially treated successfully with antipsychotics. After 6 months of remission, they were randomised for 18 months to either (a) an approach encouraging antipsychotic dose reduction or discontinuation; or (b) standard maintenance treatment: 18 months later, as expected, twice as many people had experienced a relapse in the reduction/discontinuation group as in the maintenance group. Reference Wunderink, Nienhuis, Sytema, Slooff, Knegtering and Wiersma41 However, by 7 years, the group who had been initially helped to reduce or discontinue their antipsychotics were reported to show twice the recovery rate of the maintenance patients. Reference Wunderink, Nieboer, Wiersma, Sytema and Nienhuis42
Naturalistic studies indicate that a significant minority of patients diagnosed as having psychosis and indeed schizophrenia can come off their antipsychotics without detriment in the long term. Wunderink and colleagues suggest that we can go further and try to help a larger proportion of patients to decrease or discontinue their antipsychotics. However, theirs was a small study, and it is important to emphasise that at 7 years, only 21% of patients in the dose reduction/discontinuation group had actually stopped their antipsychotics, while a similar proportion had taken a mean daily dose of less than 1 mg of haloperidol for the past 2 years. In an accompanying editorial, McGorry et al concluded that ‘It now seems probable for patients who achieve clinical remission from FEP that as many as 40% can achieve a good long term recovery with use of no or low-dose antipsychotic medication’. Reference McGorry, Alvarez-Jimenez and Killackey43
Of course, the fact that some patients remain well without antipsychotics does not mean that the drugs do not help the remainder of the patients who either stay symptomless on antipsychotics or have their symptoms ameliorated by antipsychotics.
Dopamine receptor supersensitivity
The findings of Wunderink are curious and require some explanation. Why should antipsychotics be beneficial for up to 18 months but not at 7 years? Could it be that by then, antipsychotics have lost some of their effectiveness? To examine this possibility, we need to consider the mechanism of action of antipsychotics. All antipsychotics share the same essential property – they block D2 and D3 dopamine receptors. For many years, it was thought that the abnormality in schizophrenia lay in these receptors, and therefore it seemed logical to block them. However, it is now clear that most acutely psychotic patients show excessive presynaptic synthesis of dopamine in the striatum. Reference Howes and Murray44 Thus, antipsychotics do not address the primary locus of abnormality but rather have their beneficial effects downstream of it. Is it possible that prolonged prescription of antipsychotics sometimes induces changes in the D2/D3 receptor, which diminish the effectiveness of the drugs?
Animal studies suggest that this may be the case. Reference Smith and Davis45,Reference Grace46 Thus, Samaha et al found that, with continuous administration, haloperidol and olanzapine progressively lost their efficacy in suppressing amphetamine-induced locomotion and conditioned avoidance in rats; this was reversed temporarily by a further increase in antipsychotic dose. Antipsychotic failure was linked to a 20–40% increase in post-synaptic D2/D3 receptor density. Reference Samaha, Seeman, Stewart, Rajabi and Kapur47 Ginovart et al Reference Ginovart, Wilson, Hussey, Houle and Kapur48 subsequently showed that high-dose antipsychotics lead to such an upregulation of D2/D3 receptors in cats; however, this did not occur with occupancy below 80% or with transient occupancy. Thus, D2/D3 upregulation may be a consequence of high-dose treatment. Reference Ginovart, Wilson, Hussey, Houle and Kapur48
Post-mortem studies show increased D2/D3 receptor density in the brains of people with schizophrenia; however, most of the patients in these studies had been on long-term antipsychotics. Reference Zakzanis and Hansen49 A meta-analysis of in vivo positron emission tomography (PET) and single-photon emission computed tomography (SPECT) studies Reference Howes, Kambeitz, Kim, Stahl, Slifstein and Abi-Dargham50 demonstrated that patients who had previously received antipsychotics had an increase in D2/D3 receptor availability, but, importantly, antipsychotic-naive patients were no different from controls. One PET study showed D2/D3 receptor availability greater by 30% in patients who had received long-term dopamine antagonists relative to antipsychotic-naive patients. Reference Silvestri, Seeman, Negrete, Houle, Shammi and Remington51
There have only been two longitudinal studies. Konig et al showed a 10% increase in D2/D3 receptor availability over 2 months in one patient treated with huge doses of chlorpromazine. Reference Konig, Benzer and Fritzche52 A more recent study, Reference Mizrahi, Agid, Borlido, Suridjan, Rusjan and Houle53 which scanned eight antipsychotic-naive FEP patients, showed no elevation in D2/D3 receptor availability at baseline, but after 2–3 weeks of treatment (olanzapine 10 mg/day or risperidone 2–3 mg/day) there was an increase in D2/D3 receptors in some brain regions.
The idea that antipsychotics could induce supersensitivity of the D2/D3 receptor in some patients has been championed by Chouinard and colleagues, Reference Chouinard, Jones and Annable54,Reference Chouinard and Chouinard55 who pointed out as early as 1978 that antipsychotics induce a syndrome of supersensitivity of the motor system – tardive dyskinesia – and described what they termed ‘drug-induced psychotic relapses’ associated with tardive dyskinesia after long-term high doses of FGAs. Reference Chouinard, Annable and Ross-Chouinard56 They proposed that this supersensitivity causes tolerance to antipsychotics, such that eventually they no longer control psychotic symptoms. This is supported by a follow-up study of a cohort of 8620 patients, which reported worsening of psychosis with the onset of tardive dyskinesia. Reference Tenback, van Harten, Slooff and van Os57
The evidence of antipsychotic-induced changes in the D2 receptor points to an intrinsic problem with discontinuation studies. This was noted by Carpenter and colleagues when they stated that ‘once patients are placed on medication, they are less vulnerable to relapse if maintained on neuroleptics. But what if these patients had never been treated with drugs to begin with? We raise the possibility that antipsychotic medication may make some schizophrenic patients more vulnerable to future relapse than would be the case in the natural course of the illness’. Reference Carpenter, McGlashan and Strauss58
Moncrieff, who carried out a meta-analysis of patient withdrawal studies, concluded that ‘antipsychotic discontinuation may also increase the risk of relapse over and above the risk because of the underlying disorder’. Reference Moncrieff59 Interestingly, there have been occasional reports of withdrawal psychosis occurring in people who have never had psychiatric problems but have been prescribed D2 blockers for other reasons. Reference Roy-Desruisseaux, Landry, Bocti, Tessier, Hottin and Trudel60–Reference Seeman62
Unfortunately, there are no clinical studies examining whether relapse following antipsychotic withdrawal is linked to D2/D3 receptor changes. However, Joyce, who gave haloperidol to rats for 9 months, showed that the increase in D2 receptors persisted for at least 2 months after stopping antipsychotics. Reference Joyce63 Converting rat months to their human equivalent, this would imply a period of increased vulnerability in patients following discontinuation of antipsychotics of more than a year. Reference Quinn64 Such a vulnerability period would be consistent with views expressed by patient groups such as the Hearing Voices Network. Reference Corstens, Longden, McCarthy-Jones, Waddingham and Thomas65 They remind us that that many people diagnosed as having schizophrenia have successfully come off their antipsychotics, Reference Corstens, Longden, Rydinger, Bentall and van Os66 but point out that this can be initially difficult owing to agitation, insomnia and indeed recurrence of psychotic symptoms.
The results of animal studies are clear – dopamine D2/D3 receptor upregulation and resultant supersensitivity can occur with long-term use of antipsychotics. There are some neurochemical imaging studies of patients suggesting the development of D2 supersensitivity; there is indirect evidence that this may be responsible for the decrease in efficacy of antipsychotics with their continuing prescription, and that it may contribute to relapse on stopping antipsychotics. However, there is surprisingly little longitudinal data, and methodological issues need to be addressed before a firm conclusion can be reached.
Interestingly, it seems that dopamine partial agonists do not induce D2 receptor supersensitivity in animals. Reference Tadokoro, Okamura, Sekine, Kanahara, Hashimoto and Iyo67 Furthermore, there are clinical reports that switching from a D2 blocker to the partial agonist aripiprazole worsens psychosis, as would be expected in a supersensitivity state. Reference Takaesu, Kishimoto, Murakoshi, Takahashi and Inoue68
There is an urgent need for neurochemical imaging studies addressing the question of dopamine supersensitivity in patients. We also need further RCTs to examine the risk of relapse after dose reduction and/or discontinuation of D2 blockers, as well as of partial D2 agonists, in order to establish guidelines on when and how antipsychotics can be safely reduced over time.
In contrast to patients who respond to dopamine D2 blockers, neurochemical imaging shows that a proportion of patients with TRS do not show any abnormality of dopamine synthesis. It appears that such patients have generally never responded to D2 blockers because they have nothing intrinsically wrong with their dopamine system. Reference Demjaha, Egerton, Murray, Kapur, Howes and Stone69,Reference Demjaha, Murray, McGuire, Kapur and Howes70
A second group of TRS patients have initially responded to antipsychotics but gradually lose this responsiveness. Several Japanese groups Reference Suzuki, Kanahara, Yamanaka, Takase, Kimura and Watanabe71,Reference Oda, Kanahara and Iyo72 have suggested that supersensitivity due to upregulation of D2 receptors may be the cause of this treatment resistance. Here, it is worth pondering why clozapine is the most effective treatment for patients with TRS, when it is not especially effective for patients with FEP. Reference Girgis, Phillips, Li, Li, Jiang and Wu73 Chouinard's idea that dopamine supersensitivity underlies both tardive dyskinesia and TRS reminds us that one of the most effective treatments for tardive dyskinesia is clozapine. Clozapine, of course, produces lower and more transient D2/D3 receptor occupancy than most other antipsychotics. Reference Howes, Egerton, Allan, McGuire, Stokes and Kapur74 This allows the dopamine supersensitivity of the motor system to gradually resolve, and tardive dyskinesia to slowly fade. Is it possible that the effectiveness of clozapine for some patients with TRS relies on a similar mechanism?
Some patients diagnosed with TRS do not show dopamine dysregulation, and therefore their failure to respond to D2 blockade is hardly surprising. A second group may have developed TRS secondary to prolonged D2 blockade, and the effectiveness of clozapine for such patients may rely on its ability to control positive psychotic symptoms while D2 supersensitivity resolves.
By definition, people with TRS are not going to respond to D2 blockers. However, sadly, in practice such patients are frequently prescribed high doses of D2 blockers in the misguided hope that if a moderate dose is not effective then a larger dose may be so. We urgently need studies of those people with TRS who do not respond to clozapine to ascertain whether continuing them on non-clozapine antipsychotics has some partial benefit or just produces side-effects.
What is the wise psychiatrist to do faced with the concerns we have reviewed? He or she will treat acute psychosis with the minimum necessary dose of antipsychotics, employing weight-sparing antipsychotics wherever possible; dopamine partial agonists have this property and may also be less likely to induce dopamine supersensitivity. Following recovery, the psychiatrist should work with each patient to decrease the dose to the lowest level compatible with freedom from troublesome psychotic symptoms; in a minority of patients, this level will be zero.
Such an approach requires ready access to non-pharmacological treatments. Many psychiatrists rely largely on antipsychotics, not because they dismiss the value of, for example, cognitive–behavioural therapy or vocational training, but rather because financial constraints limit the availability of such interventions. 75 Psychiatrists need to campaign to ensure that their patients have access to all beneficial treatments. The greater availability of non-pharmacological treatments would enable clinicians to determine more objectively when the balance of advantage lies in stopping antipsychotics.
Finally, it is worth recalling that Leff & Wing, who carried out the first RCT of long-term prophylaxis with antipsychotics concluded that ‘[m]aintenance therapy seems of little value in patients with a good prognosis and in the severely ill, but it is of value in the indeterminate group between these two extremes’. Reference Leff and Wing2 Of course, this historical view refers to non-clozapine antipsychotics and should be modified to suggest that the severely ill who do not respond to such drugs should be encouraged to try clozapine. Nevertheless, it is unfortunate that so little attention has been paid to this cautionary statement.