Any psychiatrist who has practiced during the last decade has probably written more than a few prescriptions for an atypical antipsychotic. But do these drugs provide any clear advantages over their predecessors? This is a question that is best answered by science, not by anecdote or by popular impression. Unfortunately, good science on this topic is hard to find. Pharmaceutical companies have exerted a tremendous influence not only over academic departments and the scientific literature, but also practitioners’ expectations. In this article we will concentrate on science, bypassing the well-publicized issues regarding infamous key opinion leaders (KOLs) and the numerous well-documented examples of fraud and deceit in the literature (which make it especially difficult to dissect out where the good science actually is).

To put it bluntly, the evidence—or rather the lack of evidence—suggests that the notion of “atypicality” has been more of a marketing concept than a pharmacological reality. The idea of atypicality arose quite early. In fact, it arose even before people realized (in the late 1980s) that clozapine (Clozaril), which had been around for more than 15 years, may have been more effective than other drugs available from the early 1970s on. The definition of “atypical” was linked initially with differences in the way these drugs affected how rats climbed up poles (eg, Costall B et al, Br J Pharmacol 1978;63(2):381P–382P). That is not a facetious remark: such simplistic behavioral tests were the common assays of the day. But they are decidedly not a reliable way of assessing drug effects—never mind that they are also far distant from what we want the drugs to do in humans suffering from psychosis.

On such grounds, drugs such as chlorprothixine (Cloxan) and thioridazine (Mellaril) were regarded as atypical at that time, but not now. Obviously, those seeking to develop new drugs more similar to clozapine hypothesized about what pharmacological properties would differentiate them from older drugs, differences that might therefore provide a lucrative advantage.

Before considering such differences, however, it is important to note that the evidence for the superiority of clozapine and other atypicals is actually quite weak: the reliability of symptom and side effect assessments is poor, and the degree of superiority is certainly small, less than a half-point on the seven-point Clinical Global Impression (CGI) scale (Lepping P et al, Br J Psychiatry 2011;198(5):341–345). Doctors may find this rather surprising, but then again, they have been inculcated with dubious and biased evidence. All of this suggests that we are dealing with a (rather expensive) castle built on sand; indeed, the general finding in medical research is that effects of small degree are of dubious significance, and frequently false (Siontis KC et al, Int J Epidemiol 2011;40(5):1280–1291).

What Does “Atypical” Mean?

Atypical antipsychotics (also called second-generation antipsychotics or SGAs) are not a class in any meaningful pharmacological sense, but the idea sure is an advertising executive’s dream come true. Various different drugs claim various different sorts of atypical features, which inevitably reminds one of Through the Looking Glass when Humpty Dumpty said, “A word means what I want it to mean, nothing more, nothing less.” Atypicality seems to have moved through several phases of pseudo-explanation—first, meso-limbic selectivity, then 5-HT2A mechanisms; now dopamine dysregulation has been run up the flagpole, but it is hard to know who is saluting it.

It gets worse. Not only are we dealing with a heterogeneous group of drugs, we are also dealing with a heterogeneous illness: not schizophrenia, but the schizophrenias. (And then there are the uses of these drugs for non-psychotic conditions like depression and anxiety, which are beyond the scope of this article.) The very real and meaningful difficulty of accommodating clinical trials to this heterogeneity has been swept regularly under the carpet, both for schizophrenia and for the non-psychotic conditions for which these drugs are used. That may be permissible—or even required—to get drugs to market, but it is disastrous for clinical science.

There are two key questions that really must be answered. The first concerns whether the evidence for the proposed theoretical basis—meso-limbic selectivity, the 5-HT2A/D2 ratio, or anything else—has been supported by time and independent replication. The second is whether drugs that exhibit those properties have been reliably and reproducibly shown to be significantly different in the ways predicted. More than two decades into this story neither of these requirements has yet been met.

Regarding notions of “atypicality,” meso-limbic selectivity has been claimed for some drugs, but PET studies in humans have failed to replicate such claims, so it is not currently possible to show whether possession of such properties (if they even exist) results in noticeable clinical differences (Kegeles LS et al, Arch Gen Psychiatry 2010;67(3):231–239). The development of selective 5-HT2A antagonists has fizzled out (Ebdrup BH et al, Expert Opinion Investig Drugs 2011;20(9):1211–1223) because of failed clinical trials, although some still support this idea. To the extent that 5-HT2A/2C/1A, or any other receptor acting as feedback modulation may influence the activity of dopamine pathways, the magnitude of change induced is likely to be small (ie, within the normal physiological range) and temporary.

Practice: Clinical Assessment

The current climate promotes and funds research geared to licensing drugs for marketing, while little funding goes towards long-term research in outcomes or head-to-head trials. As such, answers to whether SGAs provide any clear advantage over FGAs are unlikely to emerge in the near future. It is clear a huge increase in independently and publicly funded long-term research is vital.

We should also remember that the (usually short-term) assessment of drugs used for chronic disorders may not translate into meaningful long-term disease outcomes. If we are to rely on short-term drug trials, then the interim surrogate outcome measures (like subjective rating scales) must be reliably demonstrated to be related to long-term outcomes, not just symptoms over a four to six week trial. But this is simply not the case in most psychiatric research, so we must remain extremely skeptical about any presumptions concerning longer term beneficial effects or lesser side effects of these drugs. The recent reality-check concerning the minimal benefits of beta-blockers in vascular disorders is an excellent case in point, these being another group of drugs that have been in use for a number of decades (Bangalore S et al, JAMA 2012;308(13):1340–1349).

The existing evidence to support the claims for superiority of SGAs is marginal at best and dishonest at worst. Many trials remain unpublished, and effect sizes in unpublished trials are far lower than those reported in published trials (Turner EH, PLoS Med 2012;9(3):e1001189). Some observers claim there are “no important differences between any of the antipsychotics” (Kendall T, Br J Psychiatry 2011;199:266–268) and even the purported advantage of SGAs in terms of extrapyramidal side effects—including tardive dyskinesia—has been called into question (Peluso MJ et al, Br J Psychiatry 2012;200:387–392). In brief, there is no single advantage of SGAs that has been independently replicated. In fact, many recent reviews are beset with caveats about uncertainty and significance, and all of the differences are small, on the order of magnitude generated by observer errors, bias, and the sponsorship effect.

This non-superiority of the SGAs has led the British Association for Psychopharmacology (BAP) to conclude: “No double-blind trial comparing an SGA with an FGA in the acute treatment of first-episode schizophrenia has shown an efficacy advantage for the SGA, with the single exception of a head-to-head, first-line treatment trial of clozapine versus chlorpromazine conducted in China.... These results challenge the almost exclusive use of SGAs for the treatment of first-onset schizophrenia and schizoaffective disorder” (Barnes TR, J Psychopharmacol 2011;25(5):567–620). The Cochrane reviews also sing from the same sheet of music.

TCPR’s Verdict: The proposed notion of atypicality has never had sufficient evidence to support, never mind prove, any of the claims made. In fact, any existing evidence has only become weaker, not stronger, with time. After three decades, the clinical evidence of material advantages remains sparse and characterized by bias and fraud. Long-term studies are needed, but are less likely to be achieved by the research funding arrangements in place in most Western countries.

For an extended version of this article with an expanded reference list, please visit Dr Gillman’s website at www.psychotropical.com.