Schizophrenia Host Vulnerability and Risk of Metabolic Disturbances During Treatment with Antipsychotics

Harris and Barraclough (1) provided an authoritative review of historical mortality studies in schizophrenia. Their conclusion is the same as that of Allebeck (6). Moreover, from this review emerges the fact that the heightened mortality is not just due to suicide nor is it “across the board,” but there is a particularly high risk of death from cardiovascular disorders. Two recent studies confirm this impression. Miller et al. (15) conducted an extensive evaluation of causes of death among inpatients in Ohio's public mental health facilities. They reported a disproportionately higher death rate from cardiovascular disorders. Colton and Manderscheid (16) have added fuel to the fire in reporting that patients in state facilities die at ages that are 25 years younger than the average age for the general population. This number goes well beyond the previously cited rate of 15 years less life for inpatients with schizophrenia (2). Not surprisingly, this recent finding has been a “call for arms” among policy makers. Also, this finding is not merely an American phenomenon, related to our society and the high prevalence of obesity and related conditions. In a British study, Osborn et al. (4) reported a rate of stroke that was 2.5 times higher in patients with schizophrenia. Another recent publication by Saha et al. (5) showed a similar pattern of premature death in other studies from several countries and reported overall an increasing trend in cardiovascular system deaths. The reasons for the heightened risk of cardiovascular and metabolic disturbances among patients with schizophrenia are complex, and they are not uniform across patients (13, 17–19). Hennekens et al. (18) described these clearly (Table 1).

Although clinicians were well aware of the risk of agranulocytosis during clozapine therapy, the extent of weight gain and then reported diabetes mellitus during long-term therapy with clozapine caught practitioners in our field relatively unaware and has caused great concern. Allison et al. (20) showed that clozapine and other new antipsychotic medications were associated with weight gain. Henderson et al. (21) reported in their naturalistic follow-up study that one of four clozapine-treated patients had developed diabetes. There were also reports of episodes of pancreatitis and diabetic ketoacidosis—some fatal—early on in treatment with clozapine and now also with other second-generation antipsychotic (SGA) medications (22–24). Indeed, several pharmacoepidemiologic studies pointed to higher rates of diabetes mellitus in patients receiving risperidone, olanzapine, or quetiapine (13, 25–27). Ziprasidone and aripiprazole, as newer agents, appeared (at least at that time) to be less associated with this risk. The U.S. Food and Drug Administration (FDA) requested changes in the product labeling on each antipsychotic drug to reflect this risk (Table 2).

It is important to recognize that the heightened risk of metabolic disturbance during antipsychotic therapy occurs in the context of an already existing national epidemic of obesity-related problems. In the general population, especially in the United States, there has been an alarming increase in obesity and raised triglyceride and cholesterol levels (19). The term “metabolic syndrome” (MS) has been given to this constellation of disturbances of fat, insulin, and cardiovascular integrity (Table 3). There are, however, differing definitions of MS in endocrinology and epidemiology research, and the reader is referred to an excellent recent review on MS (28).

However, irrespective of the definition used, it is already evident that MS is even more prevalent is patients with schizophrenia. McEvoy et al. (29) compared the baseline rate of MS in the seminal schizophrenia Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) and found that overall 40% of patients met the criteria for MS. The rate was higher in female than in male patients (51.6% versus 36.0%). Meyer et al. (30) found no clinical differences between those with and without MS. Goff et al. (31) indicated that this rate of MS was alarmingly high and likely to increase over time. In a Belgian study that applied similar diagnostic criteria for MS, DeHert et al. (32) reported a 26.8% rate. In another 1-year study similar to CATIE, but conducted early in the course of illness, Patel et al. (personal communication, 2008) reported a 13% rate of MS even in a first- episode of schizophrenia sample.

Evidence that medications were (at least) a culprit is derived from several sources (Table 4). Newcomer (24) has provided an exhaustive review of this evidence. Nasrallah (38) provided a thoughtful synthesis of how, from a receptor-based profile, medications might contribute to the heightened risk for metabolic disturbances. However, it is not all about the medicines. In a thought-provoking study, Ryan et al. (39) reported elevated glucose levels and evidence of insulin resistance in first-episode schizophrenia patients who had never been exposed to antipsychotic medications (an important confounder is that these patients also had markedly raised cortisol levels). This finding has been confirmed. Kirkpatrick et al. (40) have offered an alternative explanation for this host vulnerability, citing evidence for widespread neurodevelopmental changes in schizophrenia that might leave the patient vulnerable to metabolic and adverse effects of medications. Mukherjee et al. (41) suggested that the host vulnerability for schizophrenia resided in a higher familial loading for diabetes. Fernandez-Egea et al. (42) have also recently replicated this finding. Moreover, Wright and Murray (43) reported a higher rate of autoantibodies associated with diabetes mellitus in patients with schizophrenia. Some have even questioned the impact of medications, citing flaws in studies that were largely retrospective in design. A recent systematic literature review by Bushe and Leonard (44) of prospective clinical trials found inconsistency in glucose levels over long-term trials. However, even this analysis of studies is incomplete and does not represent the entire breadth of studies—the majority of which at least point to some effect of medications in either causing unmasking and/or heightening the risk of metabolic disturbance.

And so the debate has raged on furiously as to the relative risk of metabolic disturbance with each of the SGAs (13). In CATIE, olanzapine was associated with the most weight gain and metabolic disturbance, whereas ziprasidone had the least (45). Aripiprazole was not included in the first two phases of CATIE. L'Italien and colleagues (46) presented data showing a greater likelihood of MS with olanzapine than with aripiprazole.

So what should I tell my patients? Well, that is a very complex problem. It does seem that the factors relevant to diabetes—e.g., smoking, family history of diabetes, and excessive weight gain—are also likely to be the contributory factors to metabolic disturbances alongside medications (17–19) (Figure 1). Nevertheless, it does also seem that antipsychotics can heighten the risk of metabolic disturbances (14, 24). It also seems that, at least on current incomplete and at times inconsistent evidence, the risk is not uniform across SGAs and some have a higher risk than others. The FDA-directed language for each of the agents is consistent with this nonuniformity (Table 1). Citrome and Stroup (47) offer an interesting way of considering this relative risk: number needed to harm (NNH). Ideally, the drug should have a large NNH. This is an indication that the risk of harm is low. Conversely, a drug should have a small number needed to treat (NNT). A low NNT indicates a robust clinical effect. In their analysis of the CATIE data, for example, Citrome and Stroup (47) found that olanzapine was the drug most likely to cause weight and metabolic disturbances in patients whereas, conversely, more patients could be treated with quetiapine before these effects were seen. Specifically, on the basis of CATIE data, they estimated that, for 100 patients treated for 18 months with either olanzapine or quetiapine, 18 more patients on olanzapine would complete treatment compared to those assigned to quetiapine. When they considered the data on weight gain, they estimated that 14 more patients receiving olanzapine would gain weight than with quetiapine. Of course, similar comparative metrics can be derived for each of the drugs. Although not without drawbacks, this is an interesting, evidence-based approach to comparing treatments (48).

There are several up-front clinical implications. To effectively manage this problem a multifaceted approach is required (Figure 2). First, it is now incumbent upon clinicians to measure weight and metabolic parameters before starting antipsychotic treatment and also to reassess these regularly during treatment. This need emerges “loud and clear” from several guidelines (49–51). However, clinicians are slow to change practice habits and at least when looked at during 2001–2003, there was a low level of measuring for MS in practice (52, 53). It is likely, especially given the medicolegal specter now surrounding this topic, that the situation has changed. How best to implement widespread change in our clinical practice is also unclear (54). Henderson et al. (55) have shown that regular monitoring of hemoglobin A1C can aid in detection of diabetes during antipsychotic therapy.

Second, we need to encourage patients to improve their nutrition and to use available weight loss programs. There is evidence that these strategies work (56). Brar et al. (57) reported that 26.5% of patients receiving a (simple) behavior intervention lost 5% or more weight. This was in comparison with weight loss seen in 10.8% of patients in regular treatment when evaluated over a duration of 14 weeks. A recent Chinese 12-week study (58) tested a behavioral modification strategy and an antiobesity drug (metformin) in first-episode schizophrenia patients who had already gained weight from their antipsychotic therapy. Patients who got standard therapy alone (i.e., no intervention for weight gain) gained more weight, whereas significant reductions in weight and in measures of insulin functioning were seen in the other three groups, i.e., those who received some intervention, either pharmacologic, nonpharmacologic, or combined medication and behavioral treatments. The best results were seen in those patients who received both metformin and the lifestyle intervention.

Third, we need to develop more comprehensive and better integrated medical models for the long-term care of patients with serious mental illness (59). Collaborations with nursing and with primary care physicians offer advantages here (60, 61). Currently, medical management for patients with serious mental illness is below par (62–64). For example, in CATIE there was less than expected prescription of medications to address the extent of diabetes and hypertension that was reported in patients (62).

In terms of medication management, switching antipsychotics is probably the best option at present. Weiden (65) provided a review of this strategy and cited several studies showing benefits such as reduction in weight and metabolic disturbances when patients switch antipsychotics. It is also important to complete the switch as incomplete switches result in polypharmacy, which itself may be associated with a greater risk for diabetes (66). Another problem with the switching strategy is that it is unclear for any individual patient what benefit or risk the patient will actually incur until the switch is made (67). Additionally, current switch studies provide limited guidance as these studies are open labeled. A federally funded switch study, Comparing Antipsychotics for Metabolic Problems (CAMP), is currently underway (36). There are also a variety of add-on strategies that have been tried to reduce weight and metabolic disturbance. Recently, DeHert et al. (68) reported benefit from adding the lipid-lowering agent rosuvastatin in patients with schizophrenia. There is also interest in considering rimonabant, a cannabinoid-1 antagonist, which has been shown to be beneficial in obese, nonpsychiatric patients (69). However, this agent is still investigational, and it also seems to have increased incidence of psychiatric side effects. The use of a histamine-1 blocker has shown some benefit in patients with schizophrenia (70). Topiramate has also been used (71). There was even interest in considering fluoxetine as an add-on to help with weight gain during olanzapine therapy (72). It should be noted, however, that even in nonpsychiatric obese populations, the evidence that these agents are helpful is mixed at best (738). Moreover, the evidence for real benefit in patients with schizophrenia is even more sparse.

The physical health of patients with schizophrenia is of great concern and requires careful and judicious treatment. It also spurs on the search for antipsychotic medications that might have a lesser metabolic side effect burden (74, 75). There is also some work examining predictors of risk for weight and metabolic disturbance. Ellingrod et al. (76) reported on a pharmacogenetic analysis suggesting that an allele of methylenetetrahydrofolate reductase might predict susceptibility for MS in patients with schizophrenia. Kaddurah-Daouk et al. (77) have shown how metabolomics—a new application of genetics and informatics—can assist in predicting risk. In the meantime, while we are waiting for such research approaches to be confirmed in later studies and mature out into practice, treatment decisions will need to be individualized, and clinicians will need to continue to rely on their judgment, careful assessment, and continued learning from the literature as they engage in complex decision making (13).