Jack D. McCue, MD Clinical Professor of Medicine, University of California at San Francisco.

The second-generation antipsychotic drugs are usually referred to as "atypical" antipsychotics because the first drug in this class, clozapine, was less likely to cause movement disorders—the most dreaded complication of the first-generation drugs (or "typical" antipsychotics), such as haloperidol or the venerable chlorpromazine. Clozapine, however, was difficult to use, not only because it caused troublesome side effects but also because 1% to 2% of patients developed potentially life-threatening agranulocytosis in the first 6 months of therapy, requiring weekly white blood cell counts initially, then monthly counts after the first 6 months. Pharmaceutical companies attempted to design drugs that would capture what appeared to be the greater antipsychotic efficacy of clozapine, but without its toxicity. These drugs—olanzapine, quetiapine, risperidone, and ziprasidone—are indeed effective, do lack the problematic hematologic toxicity of clozapine, and are associated with a much reduced incidence of movement disorders, the most disfiguring being tardive dyskinesia. As a result, they have essentially displaced the first-generation antipsychotics in the treatment of schizophrenia. They have, as well, displaced the first-generation antipsychotics in a common and controversial off-label use as sedatives for agitation in dementia.

Concerns initially voiced in 2001 about the possibility of metabolic complications of the second-generation antipsychotics were eventually validated by a series of retrospective studies that investigated the benefits of these drugs. In 2005, Lieberman and colleagues¹ reported the comprehensive prospective trial sponsored by the National Institute of Mental Health known as the CATIE (Clinical Antipsychotic Trials of Intervention Effectiveness) study. This was a double-blind, randomized trial of 1493 patients with chronic schizophrenia in the maintenance phase of treatment who were randomly assigned to receive the first-generation antipsychotic perphenazine or 1 of the 4 second-generation agents. Olanzapine was judged to be more effective than the other drugs in 3 rather indirect measurements—time to discontinuation of treatment for any cause over 18 months, worsening of patients who were switched from olanzapine as a prestudy baseline drug to one of the other study drugs, and greatest improvement after starting olanzapine after treatment failure with one of the other drugs. In fact, one could more securely assert that the study demonstrated that olanzapine was the best tolerated, although not necessarily the most effective, of these drugs. The CATIE study did, however, appear to confirm the conclusion of an earlier meta-analysis by Davis and colleagues² that olanzapine is more effective than the comparator antipsychotics. Thus the literature supports the opinion that although none of the second-generation antipsychotics is necessarily the preferred choice, olanzapine has been shown to be marginally more effective in major comparative trials.

Alas, there are no more "free lunches," even in San Francisco. ("Free lunch" was derived from an American saloon custom in the later nineteenth-century, memorably described by Rudyard Kipling, who in a late 1880s visit to San Francisco "came upon a barroom full of bad Salon pictures in which men with hats on the backs of their heads were wolfing food from a counter. It was the institution of the ?free lunch' I had struck. You paid for a drink and got as much as you wanted to eat....Remember this if ever you are stranded in these parts."³ Because the lunch cost the owner more than the price of a drink, saloon owners expected customers to purchase multiple drinks, as was the custom in that heavy-drinking era. It was a classic risk?benefit calculation by the owners. Whether the customers developed metabolic complications is not recorded.)

The prospective CATIE trial also showed that 30% of the olanzapine patients gained more than 7% of their body weight compared with only 7% to 16% of the patients using the other drugs.¹ In a large study also referenced by Iqbal and Thomas in their article in this issue of the journal, Lambert et al⁴ found similarly elevated risks of new-onset diabetes mellitus type 2 in 15,767 veterans treated with olanzapine, quetiapine, or risperidone (hazard ratio, 1.64, 1.60, 1.67, respectively) compared with haloperidol as the referent drug. The incidence of diabetes in this 1-year study ranged from 2.0 per 100 person-years for haloperidol to 3.6 per 100 person-years for quetiapine—or about a 60% to 70% increase for the second-generation antipsychotics compared with haloperidol.⁴ Overall, one third of the new cases of diabetes in these veterans could be attributed to their antipsychotic therapy.

Although previous studies have stronger statistical validity than the relatively small, shorter study of Iqbal and Thomas (which included 977 patients who were followed for 15 weeks), their study does put a practical face on the cost?benefit dilemma faced by physicians, who must choose an antipsychotic medication for initiation or continuation of schizophrenia therapy. During the short period of time of Iqbal and Thomas' study, 6 patients—nearly 1%—developed diabetes (0.61%, with an average time of onset of 8 weeks). Overweight African Americans are at greater risk of developing diabetes than similarly overweight whites, and 83% of their predominantly African-American patients who developed diabetes were overweight at the onset of treatment (average body mass index [BMI], 29.97 kg/m²—borderline for obesity); all those who developed diabetes gained weight, and 83% increased their BMI by more than 5%. Iqbal and Thomas found that, as has been noted in the clinical trials they cite in their discussion, olanzapine had a significantly higher risk of new-onset diabetes compared with other second-generation antipsychotics. The number of patients in Iqbal and Thomas' study, however, is too small to be clinically meaningful if the excess risk had not already been observed in previous, larger clinical trials.

Some messages to take away from Iqbal and Thomas' study: First, weight gain is common with the institution of antipsychotic therapy, and may be greatest with olanzapine (too many big lunches?). Second, the association of the second-generation antipsychotics with diabetes is real, can emerge very quickly, and can cause symptomatic diabetes (or even ketoacidosis). Although diabetes tends to develop in patients who have significant weight gain (and similarly, those who develop diabetes tend to have gained more weight than those who do not develop diabetes), the risk is not entirely attributable to the effect of weight gain. Third, the association of diabetes and the second-generation drugs is clinically prominent enough to be encountered by clinicians, not just in epidemiologic studies. Fourth, although monitoring fasting blood sugars (the role of hemoglobin A_1c monitoring has not been clarified, but could be useful after the first few months of therapy) is most important in patients who are already obese and who experience additional weight gain, it is not clear whether monitoring alone in that subgroup is safe. Fifth, the diabetes is usually well-controlled with oral agents, although some patients do require insulin.

Finally, a caution from the CATIE trial.¹ Patients who are doing well with olanzapine should not have their regimen altered only because of the greater risk of diabetes—the risk overall is low, and there is a significant risk that their schizophrenia symptoms may worsen if they are switched to a different second-generation antipsychotic.¹ Nor should physicians avoid starting patients on olanzapine therapy just because of concerns about metabolic side effects. But for those at highest risk for diabetes or those who have other metabolic risk factors (eg, hyperlipidemia), serious consideration should at least be given to initiation of therapy with one of the other second-generation drugs, such as risperidone, which may have the lowest risk of inducing diabetes (excluding clozapine, which has otherwise unacceptable risks).⁵

Aripiprazole may have the lowest risk of inducing or worsening hyperlipidemia. Similarly, it is doubtful whether discontinuation of second-generation antipsychotics would benefit patients who develop diabetes—the weight of evidence (sorry) would suggest not. Beginning therapy with metformin, as is the common initial choice in type 2 diabetes, is the best course of action.

References

1. Lieberman JA, Stroup TS, McEvoy JP, et al, for the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) investigators. Effectiveness of antipsychotic drugs in patients with chronic schizophrenia. N Engl J Med. 2005;353:1209-1223.
2. Davis JM, Chen N, Glick ID. A meta-analysis of the efficacy of second-generation antipsychotics. Arch Gen Psychiatry. 2003;60:553-564.
3. Kipling R. In San Francisco. Reprinted in Miller J, ed. San Francisco Stories: Great Writers on the City. San Francisco, Calif: Chronicle Books; 1990:123-142.
4. Lambert BL, Cunningham FE, Miller DR, et al. Diabetes risk associated with use of olanzapine, quetiapine, and risperidone in Veterans Health Administration patients with schizophrenia. Am J Epidemiol. 2006;164:672-681.
5. Ramaswamy K, Masand PS, Nasrallah HA. Do certain atypical antipsychotics increase the risk of diabetes? A critical review of 17 pharmacoepidemiologic studies. Ann Clin Psychiatry. 2006;18:183-194.