The role of atypical antipsychotic drugs in schizophrenia

This issue of the Bulletin contains a series of thought-provoking articles concerning the choice of antipsychotics in schizophrenia (Adams & Gilbody, this issue; Bebbington, this issue; Healy, this issue; Hogman, this issue; Mortimer, this issue). In other words, atypicals or not and if so when. This is an argument that has gone on since the reintroduction of clozapine in 1990 (Kerwin, 1996). I will not, of course present my own views in this editorial, rather try and set the scene.

The atypical antipsychotic market-place is becoming overcrowded and therefore frenetically complicated by an ever increasing clamour and counter-clamour of vested interests. Two to three more new drugs in this class are to enter the scene in the next few years. There are widely divisive opinions and widely differing practices in the use of novel antipsychotics. In the US atypical antipsychotic prescribing now far outweighs typical antipsychotic prescribing, with risperidone and olanzapine being the most popular drugs. In the UK only a minority of patients are receiving atypical drugs and there is widespread poor prescribing of antipsychotics that will need to be corrected before any implementation of guidelines can be instituted (Taylor et al, 2000).

The open question in this debate is where lies the point of clinical equipoise? In other words, where do the genuine uncertainties lie that demand investigation and should these be explored by randomised controlled trials (RCT) or naturalistic trials? This point of equipoise has moved over the years. Clozapine was hailed as a wonder drug completely free of neurological side-effects in 1990 and each successive new drug was also fanfared. However, a wider examination of the pivotal trials makes it clear that new drug efficacy is equal to older drugs and whereas the traditional side-effect profile is superior, a differing side-effect profile may pose equal difficulties in tolerability for new drugs. So the points of clinical uncertainty are a moving feast. In addition, with the eschewing of high dose antipsychotics we are learning more about how to use low dose strategies for conventional antipsychotics. There are several difficulties in untangling these dilemmas as I see them. There is an argument for more naturalistic studies using hard endpoints. This is technically correct but the size of the sample needed is enormous. In addition, we can't do without the smaller RCT as it would be wasteful to embark on natural studies without first ascertaining baseline efficacy under controlled conditions. No prudent drug company would be expected to do this. Furthermore, if the RCT data are good, should patients be made to wait further years before the naturalistic data come in? What happens if further new generations of superior drugs come in and leapfrog the atypicals before the naturalistic data are known? From a clinical and patient point of view this seems a recipe for stagnation. I strongly believe that all these arguments will be worthless unless we consider this debate to be aimed at our incident cohort of patients rather than our prevalent cohort. Multi-episode patients currently under treatment are in the main partially treated or difficult to switch, therefore undermining any rigid guidelines. First-episode patients are more treatment responsive and more side-effect sensitive (Lieberman et al, 1993). Therefore, whatever the pros and cons of different drug classes, they are more likely to be amplified at first episode and in turn such patients are more likely to achieve a better prognosis with the right choice. There are other factors too. What about the duration of untreated psychosis? If it really is related to prognosis (Haas et al, 1998) does the current incident cohort have time to wait for the outcome of naturalistic trials or the deliberations of the National Institute for Clinical Excellence (NICE)? What is NICE going to do? I can't see it settling any arguments, because there are few head to head studies comparing new drugs and in essence no clinical effectiveness data because there are no naturalistic trials. Cost-effectiveness will have to be modelled and that will rely on RCTs. I think NICE will only be able to pronounce on these drugs as a class. In which case why not go for the cheapest, e.g. zotepine, a drug with virtually no marketing muscle behind it?

It will not be long before we have a third wave of neuroleptics, with novel mechanisms of action such as partial dopamine agonists, selective D3 blockers and more. How can we have progress with the use of these drugs unless we get a move on with deciding about what, after all, are now established treatments in schizophrenia. Hopefully the debate in this issue will help to facilitate this important discussion.