The potential contribution of antipsychotic medication to the risk of diabetes in patients with severe mental illness has received significant attention in the psychiatric literature (1). In a recent review in Diabetes Care, Lean and Pajonk (2) claim to have summarized the “evidence for a causal link” between certain atypical antipsychotics and diabetes. However, their argument relies largely on spontaneous reports and retrospective, or uncontrolled, studies that are not suited to address causation. Furthermore, recent controlled prospective studies raise doubts about the authors’ suggestion of a “direct metabolic effect” for drugs such as olanzapine.
The authors use case reports to argue for differential “diabetogenic potential of the atypical antipsychotics ” (in Table 1 of their article) and claim that risperidone “appears to have the least propensity to cause diabetes.” They did not cite studies by Koller et al. (3) and Hedenmalm et al. (4) documenting additional cases of diabetes, including cases of diabetic ketoacidosis and death, in patients taking risperidone. Neither did they point out that spontaneous reporting of adverse events is poorly suited to establishing disease incidence (5) and thus relative risk. Three pharmacoepidemiology studies are referenced in the article (6–8) that report lower diabetes risk among risperidone users. Other studies in which risperidone use was associated with a higher risk are not discussed (9–11). Perhaps more importantly, none of the studies directly comparing patients treated with different atypical antipsychotics reported statistically significant differences in rates of new diabetes, suggesting that diabetes risk is comparable among the various agents represented in these studies.
In discussing potential mechanisms of drug-induced diabetes, Lean and Pajonk reference studies by Cagliero (12) and Newcomer (13) as evidence that olanzapine and clozapine adversely affect glucose metabolism and insulin sensitivity. Such conclusions are weakened by the fact that both studies were cross-sectional and without randomization to treatment. Using the hyperinsulinemic-euglycemic clamp method in a randomized prospective clinical trial (14), we found no significant differences in insulin sensitivity in normal individuals treated with olanzapine, risperidone, or placebo. A study in schizophrenia patients likewise did not show differences in insulin sensitivity comparing olanzapine, risperidone, and typical antipsychotic treatment cohorts (15). To assess changes in glycemic control during antipsychotic treatment, we analyzed our own clinical trial database (>5,000 patients in randomized controlled studies) and have not seen significant differences in the rates of new diabetes when olanzapine cohorts were compared with placebo or haloperidol or risperidone treatment groups (16) (Lilly data on file). Likewise, there were no significant differences among these groups in the number of patients exceeding potentially meaningful cutoffs in casual blood glucose values (e.g., ≥126, 140, 160, or 200 mg/dl) (17).
The article by Lean and Pajonk does make the important point that patients with severe mental illnesses such as schizophrenia are at increased risk for diabetes and that this association predated the introduction of newer antipsychotic drugs. Higher diabetes prevalence, irrespective of antipsychotic treatment choice, was also seen in the study by Sernyak et al., which they cite (6). According to Lean and Pajonk, this study showed that “there was a significant association between the development of diabetes and prescription of quetiapine, clozapine, and olanzapine but not risperidone.” They fail to report that this was only true when the atypical antipsychotic cohorts were individually compared with the cohort taking older “typical” antipsychotics. In fact, Sernyak et al. report that the prevalence of diabetes among younger antipsychotic users (typical and atypical) was over five times the expected rate in the general population. Because risperidone users made up >40% of this group and the odds ratios for diabetes were not significantly different between risperidone and the other atypical agents (based on overlapping CIs), it seems clear that patients treated with risperidone are not protected from developing diabetes. The important point is that this group of patients represents a high-risk group regardless of the antipsychotic used. This risk is related, at least in part, to characteristics inherent in this population (16). Whether the medications add to this underlying risk has not, in our opinion, been established.
Serious mental illnesses, such as schizophrenia, are debilitating and difficult to treat. Clinicians need an accurate picture of the relative risks and benefits of available treatment options. Unfortunately, the incomplete presentation of the data by Lean and Pajonk will only worsen the confusion. Additional, well-designed research is needed to provide the clarity and confidence clinicians and their patients deserve.
References
Olanzapine is marketed by Eli Lilly.
