Abstract
Background:
Antipsychotic-related weight gain and metabolic adverse effects have become a major focus, especially in youth.
Methods:
Review of randomized, cohort, and pharmacoepidemiologic studies of antipsychotic-related weight gain and metabolic adverse effects and of interventions for their reduction in youth.
Results:
Across 34 published head-to-head and placebo-controlled studies in youth with psychotic and bipolar disorders, weight gain ranged from 3.8 to 16.2 kg with olanzapine (n=353), 0.9–9.5 kg with clozapine (n=97), 1.9–7.2 kg with risperidone (n=571), 2.3–6.1 kg with quetiapine (n=133), and 0–4.4 kg with aripiprazole (n=451). In 24 placebo-controlled trials, the numbers-needed-to-harm for weight gain ≥7% in youth with bipolar disorder and schizophrenia were 39 (confidence interval [CI]: −1 to +6, not significant) for aripiprazole, 36 (CI: −1 to +7, not significant) for ziprasidone, 9 (CI: 7–14) for quetiapine, 6 (CI: 5–8) for risperidone, and 3 (CI: 3–4) for olanzapine. Data in youth with autism and disruptive behavior disorders, available only for some antipsychotics, suggest greater weight gain, possibly due to less prior antipsychotic exposure. Three-month results from a large cohort study in antipsychotic-naïve youth indicated that metabolic effects differ among second-generation antipsychotics, despite significant weight gain with all studied agents, suggesting additional, weight-independent effects. Further, pharmacoepidemiologic work indicates that antipsychotic polypharmacy increases the risk for obesity (odds ratio [OR]: 2.28 [CI: 1.49–3.65]) or any cardiovascular, cerebrovascular, or hypertensive adverse event (OR: 1.72 [CI: 1.10–2.69]). However, despite marked weight gain and its greater impact on youth, monitoring rates are low and studies of pharmacologic and behavioral interventions are extremely limited.
Conclusions:
More research is needed to develop strategies to minimize antipsychotic-related weight gain and metabolic effects in youth and to discover treatments with lower risk potential.
Introduction
Although SGAs provide significant benefits for children and adolescents with a variety of severe psychiatric conditions (Correll et al. 2011a), including schizophrenia (Kumra et al. 2008b), bipolar mania (Chang 2008), and aggression associated with autistic disorder and disruptive behavior disorders (Jensen et al. 2007; McDougle et al. 2008), the development of rapid and large weight gain, as well as metabolic adverse effects have caused considerable concern (Correll 2008; Sikich et al. 2008; Correll et al. 2009). Until new, safer and effective remedies are developed, the understanding of the risks of individual antipsychotics and of the best ways to monitor and manage these important side effects is an essential part of psychiatric practice.
In this article, we reviewed the relative risks of antipsychotics for age-inappropriate weight gain and adverse metabolic changes in children and adolescents. We performed a PubMed search for primary studies and review articles, using the following search terms: “child, children, childhood, adolescent, adolescence, youth, pediatric” and “weight, body mass index (BMI), metabolic, cardiometabolic, glucose, hyperglycemia, diabetes insulin, lipid, lipids, hypercholesterolemia, hypertriglyceridemia, hyperlipidemia, dyslipidemia, metabolic syndrome.” After identifying the published literature relevant for the topic of this article, we performed a selective review of the evidence, focusing on quantitative estimates of weight gain and metabolic risk of antipsychotics in youth. In this review, we did not focus on the potential for cardiac risks related to antipsychotics, such as QTc prolongation, arrhythmia, and sudden death (Correll et al. 2011c; De Hert et al., 2011). Finally, we outlined gaps in the literature, indicating areas for further research.
Consequences of Age-Inappropriate Weight Gain and Metabolic Abnormalities During Childhood
As the potential benefits of SGAs have been demonstrated by recent research, so have been some of their adverse effects. It appears that the tradeoff between the lower levels of dopaminergic antagonism, which is associated with less of the previously much feared extrapyramidal side effects (Correll 2008), for more antagonism at serotonin, histamine, and other receptors, has given SGAs a greater propensity to cause weight gain and metabolic derangement compared with first-generation antipsychotics (FGAs). This was broadly demonstrated in adults in a 1999 review of FGAs and SGAs (Allison et al. 1999). Subsequently, adverse effects of antipsychotics on glucose and lipid metabolism have also been published in adults (Newcomer 2005). Nevertheless, there is growing recognition that the polarized dichotomization of FGAs versus SGAs cannot be upheld. Rather, the risk potential for weight gain and metabolic abnormalities is heterogeneous in both antipsychotic classes (Correll et al. 2009; Kane and Correll 2010; Nielsen et al. 2010).
Cardiometabolic adverse effects are particularly concerning, as obesity and abnormalities in glucose and lipid metabolism are related to the development of type 2 diabetes, metabolic syndrome, and cardiovascular morbidity and mortality (Barness et al. 2007). The relationship between overweight, obesity, and metabolic abnormalities with poor cardiovascular outcomes might even be accelerated or accentuated when age-inappropriate weight gain and metabolic abnormalities begin during development (Nathan and Moran 2008).
In the general community, almost one-third of U.S. children are at least overweight (i.e., above the 85th percentile of BMI for sex and age), one-sixth are obese (i.e., at or above the 95th BMI percentile), and more than 1 out of 10 youth are severely obese (i.e., at or above the 97th BMI percentile), with no sign of decreasing prevalence rates (Ogden et al. 2008), predicting a public health problem of considerable personal and societal costs in the coming decades (Cali and Caprio 2008; Ogden et al. 2010; Huang et al. 2009). The medical severity of pediatric obesity is becoming increasingly evident, with longitudinal data from large population-based studies showing that carotid intimal media thickness in adults, a risk factor for stroke, is related to low-density lipoprotein (LDL) cholesterol levels (Davis et al. 2001) and high BMI (Li et al. 2003) in childhood. Some of the cardiometabolic effects of obesity are even becoming manifest in children, supported by increased rates of metabolic syndrome (Weiss et al. 2004) and carotid intimal media thickening in youth (Urbina et al. 2009). Pharmacoepidemiologic work in pediatric obesity examining rates of pediatric use of antihypertensive, antidiabetic, and antidyslipdemic medications has also shown increases in the last several years, suggesting higher rates of hyperglycemia and cardiovascular disease in childhood. Type 2 diabetes mellitus (T2DM), once a disease almost exclusively found in adults, is increasingly prevalent in children both in the United States (Pinhas-Hamiel et al. 1996; Rosenbloom 2000; Molnar 2004; Jerrell and McIntyre 2008) and in the rest of the world (Freeman-Fobbs 2003). Moreover, there are also negative effects of cardiometabolic abnormalities on mental outcomes. For example, pediatric obesity has been correlated with deficits in orbitofrontal cortical gray matter and cognitive functioning (Maayan et al. 2011), and type 2 diabetes has been associated with additional structural white matter brain impairments (Yau et al. 2010).
Comparative Effects of Antipsychotics on Body Composition in Youth
Antipsychotic-associated weight gain is all the more concerning when viewed against the backdrop of the data on the proximal and distal adverse effects of pediatric obesity. Nevertheless, the cardiometabolic risk is not equal across medications and populations, as shown in adults (American Diabetes Association et al. 2004; Lieberman et al. 2005; Newcomer 2005; Kahn et al. 2008).
A review of antipsychotic cardiometabolic effects in pediatric and adult studies has shown that youth may be even more vulnerable to weight gain than adults (Correll and Carlson 2006; Correll 2008). To assess the age-dependent weight gain liability of risperidone, a systematic review was performed showing increased weight gain in youth (Safer 2004). This was followed by a more recent meta-analysis of over 3,000 subjects in pediatric and adult trials also involving risperidone, showing overall equal all-cause and specific-cause discontinuation rates and even slightly higher efficacy in children, but over three times the percentage of body weight gain in youth compared with baseline than in adults (Maayan et al. 2010b). The higher acute weight and metabolic burden of SGAs in youth compared with adults has also been demonstrated with olanzapine (Kryzhanovskaya et al. 2009a), and with several other SGAs (Correll et al. 2010).
However, despite this potentially increased cardiometabolic side effect burden in pediatric patients, clinical studies directly comparing weight gain and metabolic effects of different antipsychotics in youth are scarce. The available evidence comes from four sources: (1) randomized controlled trials with an active comparator, (2) indirect comparison of placebo-controlled trials, (3) naturalistic data collected under standardized conditions, and (4) pharmacoepidemiologic studies (Table 1).
Note: only studies that reported relevant results are included in this table.
SD=standard deviation; SCHIZ=schizophrenia; CLZ=clozapine; HAL=haloperidol; OLZ=olanzapine; RIS=risperidone; MOL=Molindone; ARI=Aripiprazole; PBO=placebo; QTP=quetiapine; PDD=pervasive developmental disorder; ZIP=ziprasidone; Behav=behavioral disorder; N/A=not available; NS=not significant; BMI=body mass index; OR=odds ratio; FDA=Food and Drug Administration;ADHD=attention-deficit/hyperactivity disorder; w=weight; wks=weeks; mon=months; BMI%=BMI percentile.
Active comparator trials with two or more SGAs head to head
An early active comparator study comparing efficacy and side effects in early onset schizophrenia examined 26 patients randomized to either olanzapine or clozapine for 8 weeks. The clozapine group had a somewhat better response, particularly in terms of negative symptoms, and both groups showed weight gain of approximately 4 kg (Shaw et al. 2006). A 12-week pediatric study examining treatment-resistant schizophrenia in 39 youth showed again superior efficacy for clozapine and approximately equal levels of weight gain with 3 participants on clozapine and 2 on olanzapine gaining at least 7% of their body weight (Kumra et al. 2008a). The results of this study, while clinically important, make the relative orexigenic contribution of clozapine and olanzapine difficult to interpret as all subjects by definition had been on prior antipsychotic.
A larger study comparing 50 psychotic youth treated with risperidone, olanzapine, and haloperidol reported significant weight and BMI gain in all three groups, with olanzapine (7.1 [±4.1] kg and 2.3 [±1.2] kg/m2) causing the most weight gain, followed by risperidone (4.9 [±3.6] kg; and 1.4 [±1.2] kg/m2) and then by haloperidol (3.5 [±3.7] kg and 1.1 [±1.2] kg/m2). While there was a trend toward risperidone and olanzapine causing a more rapid rate of weight gain than haloperidol, this difference did not reach significance. Other randomized open-label research includes three studies with quetiapine. In one 6-month study, for example, psychotic adolescents on olanzapine gained nearly three times more weight (15.5 kg) than those on quetiapine (5.4 kg) (Arango et al. 2009). Another study comparing quetiapine, olanzapine, and risperidone in youth with psychosis showed a similar percentage of participants on quetiapine and olanzapine to gain at least 7% of their baseline body weight (55.6% and 60%, respectively); however, more patients on risperidone (80%) gained at least 7% weight (Jensen et al. 2008). A more recent, small study directly comparing risperidone and quetiapine found more participants on quetiapine to gain at least 5% and 10% of their baseline body weight than those on risperidone (Swadi et al. 2010). Possibly due to the small sample size in these studies, none of the comparisons between quetiapine and risperidone was statistically significant (Jensen et al. 2008; Swadi et al. 2010).
One of the most rigorous comparative antipsychotic pediatric trials was the National Institute of Mental Health (NIMH)–sponsored Treatment of Early Onset Schizophrenia Spectrum Disorders (TEOSS) study. This was a study of 119 children randomized to olanzapine, risperidone, or the FGA molindone (coadministered with double-blind, placebo-controlled benztropine 0.5 mg two times per day). Efficacy in the blinded 8-week trial was equivalent between the groups. However, weight gain and adverse metabolic effects differed significantly. Youth gained an average of 6.1 kg on olanzapine, 3.6 kg on risperidone, and 0.3 kg on molindone. There were also significant differences in insulin and some lipid parameter levels, disfavoring olanzapine, leading the NIMH Data Safety Monitoring Board DSMB to halt recruitment into the olanzapine arm after an interim analysis. Subjects who completed the acute, 8-week TEOSS study could be followed in a 44-week, open label, extension study. During the extension study (Findling et al. 2010b), minimal differences in cholesterol and insulin emerged, but there was a dramatic weight increase in the molindone group, bringing it on par with the other two agents. In analyzing the surprising lack of body weight and metabolic risk difference at the end of the 1-year study, the authors point out the potential role that tolerability had in the continuation of individuals, particularly in the 44-week extension, in that youth most vulnerable to weight gain on olanzapine or risperidone might have dropped out earlier in the study, leading to an underestimation of the total weight gain potential.
Placebo-controlled trials
Comparing data from placebo-controlled trials allows a broader group of SGAs to be contrasted; however, this advantage is tempered by the potential confounds that are introduced by comparing across different study populations and study methodologies. With aripiprazole, five placebo-controlled trials have been completed (Findling et al. 2008b,2009; Marcus et al. 2009; Owen et al. 2009; Tramontina et al. 2009), three of them showed increased weight gain over placebo. With olanzapine (Hollander et al. 2006; Tohen et al. 2007; Kryzhanovskaya et al. 2009a) and quetiapine (Food and Drug Administration 2010; DelBello et al. 2007b, 2009), three placebo-controlled trials have been completed, all showing significant weight gain over placebo.
Risperidone has been the most widely studied antipsychotic in youth with 14 published placebo-controlled trials in youth with autism, conduct disorder, bipolar disorder, and schizophrenia (Findling et al. 2000; Buitelaar et al. 2001; Hellings et al. 2001; Aman et al. 2002, 2005; Snyder et al. 2002; Scahill et al. 2003; Shea et al. 2004; Luby et al. 2006; Nagaraj et al. 2006; Armenteros et al. 2007; Haas et al. 2009a, 2009c). All studies showed significant weight gain over placebo with the exception of a 13-subject study that was likely underpowered (Van Bellinghen and De Troch 2001). With ziprasidone (Food and Drug Administration 2009; Sallee et al. 2000; DelBello et al. 2008a; Findling et al. 2010a) four placebo-controlled studies have been completed, none of them showing significant weight gain over placebo. To compare the relative tolerability, efficacy and safety of SGAs in youth with psychotic and bipolar disorders, Fraguas and colleagues performed a comprehensive review of 34 published head-to-head and placebo-controlled studies lasting between 3 weeks and 12 months including 2,719 youth (Fraguas et al. 2011). Mean weight gain ranged from 3.8 to 16.2 kg with olanzapine (n=353), from 0.9 to 9.5 kg with clozapine (n=97), from 1.9 to 7.2 kg with risperidone (n=571), from 2.3 to 6.1 kg with quetiapine (n=133), and from 0 to 4.4 kg with aripiprazole (n=451). Owing to the significant heterogeneity between studies, the authors felt unable to make a strict ranking of relative orexigenicity between SGAs.
In a more recent systematic review of placebo-controlled studies of antipsychotics in youth independent of diagnosis, De Hert and colleagues were able to abstract weight gain data from 24 trials with a total of 3,048 patients (De Hert et al. 2011). Across these studies, ziprasidone was associated with the lowest weight gain (−0.04 [95% confidence interval [CI]: −0.38 to +0.30] kg), aripiprazole (0.79 [CI: 0.54 to 1.04] kg), quetiapine (1.43 [CI: 1.17 to 1.69] kg), and risperidone (1.76 [CI: 1.27 to 2.25] kg) were intermediate, and olanzapine was associated with weight gain the most (3.45 [CI: 2.93 to 3.97] kg). Using ≥7% weight gain as clinically significant, the authors further calculated numbers-needed-to-harm (NNH) in these placebo-controlled studies, finding NNHs in the entire sample of 36 (CI: −1 to +7, not significant) for ziprasidone, 12 (CI: 9–17) for aripiprazole, 9 (CI: 7–14) for quetiapine, 6 (CI: 5–7) for risperidone, and 3 (CI: 3–4) for olanzapine, with lower NNHs indicating stronger weight gain potential. However, looking at potential factors for the heterogeneity of the results, it appeared that significant weight gain was more prevalent in youth with autistic disorder who were also younger and had likely experienced less prior antipsychotic weight gain. Therefore, NNHs were also calculated for patient cohorts excluding autistic youth who were not part of each of the placebo-controlled trial data base. In these more comparable samples, the respective NNHs for weight gain of ≥7% were 39 (CI: −1 to +6, not significant) for aripiprazole, 36 (CI: −1 to +7, not significant) for ziprasidone, 9 (CI: 7–14) for quetiapine, 6 (CI: 5–8) for risperidone, and 3 (CI: 3–4) for olanzapine.
Based on these figures, the few head-to-head studies comparing clozapine and olanzapine, and bearing in mind that direct head-to-head studies are needed to confirm the results from these indirect comparisons of placebo-controlled trials, the weight gain potential ranks as follows: olanzapine ≥clozapine >risperidone ≥quetiapine >aripiprazole=ziprasidone.
As noted above, children and adolescents appear more susceptible to this side effect. This could be because of developmental differences, making them more vulnerable to the orexigenic effects of SGAs or to the fact that they are more likely to be antipsychotic naïve. A recent meta-analysis looking at only studies involving hospitalized antipsychotic-naïve adults, including also some individuals as young as 15 years, reported a mean weight 3.8 kg across trials (Tarricone et al. 2010). None of the studies used the relative low-weight-gain agents aripiprazole or ziprasidone. This amount of gain is not specified by agent and certainly is less than the >6 kg seen with olanzapine in the gold-standard TEOSS trial, suggesting that age may be a factor as well. Moreover, when comparing weight change results across the age range, it is important to recognize that children and adolescents generally have lower baseline body weight. Therefore, even similar weight increase over short study durations of 3 months or less (i.e., when growth plays a rather minimal role for weight change) represents greater abnormal weight gain in youth relative to their body weight.
Naturalistic cohort studies
To date, the most comprehensive prospective comparative data regarding antipsychotic-naïve youth come from the Second-Generation Antipsychotic Treatment Indications, Effectiveness and Tolerability in Youth (SATIETY) study, a naturalistic study of children and adolescents aged 4–19 years started on antipsychotics (Correll et al. 2009). Initial results from the cohort of 272 antipsychotic-naïve youth demonstrated significant weight gain over the first 3 months of treatment with all SGAs, including aripiprazole: 4.4 kg, 58.4% gaining ≥7% weight; risperidone: 5.3 kg, 64.4% gaining ≥7% weight; quetiapine: 6.1 kg, 55.6% gaining ≥7% weight; and olanzapine: 8.5 kg, 84.4% gaining ≥7% weight. By contrast, the small group of psychiatrically ill youth who refused or stopped SGA treatment within <4 weeks gained only 0.2 kg, 0% gaining ≥7% weight (Correll et al. 2009).
Pharmacoepidemiologic studies
Using data from Medicare in South Carolina, McIntyre and Jerrell compared 4,140 children and adolescents newly prescribed one of six SGAs or two FGAs between January 1998 and December 2005, with 4,500 children and adolescents not on antipsychotics (McIntyre and Jerrell 2008). The authors reported that children on antipsychotic were more likely to have a chart diagnosis of obesity (odds ratio [OR]: 2.13 (1.85 to 2.5) than the control population (McIntyre and Jerrell 2008). While no individual antipsychotic was associated with increased rates of obesity, the following factors were associated with increased obesity rates: multiple antipsychotics (2.28 [1.49–3.65]) (McIntyre and Jerrell 2008), female gender (1.77 [1.43 to 2.17]), African-American race (1.45 [1.18 to 1.79]), and being older than age 13 (1.34 [1.08 to 1.66]) (Jerrell 2010).
Comparative Effects of Antipsychotics on Glucose and Lipid Metabolism
While the amount of developmentally inappropriate weight gain is an important risk factor for later endocrine and cardiovascular illness, humoral measures such as insulin resistance and cholesterol, and triglyceride abnormalities are more closely linked to these sequellae (Table 2).
SD=standard deviation; SCHIZ=schizophrenia; CLZ=clozapine; HAL=haloperidol; OLZ=olanzapine; RIS=risperidone; MOL=Molindone; ARI=Aripiprazole; PBO=placebo; QTP=quetiapine; PDD=pervasive developmental disorder; ZIP=ziprasidone; Behav=behavioral disorder; N/A=not available; NS=not significant; BMI=body mass index; OR=odds ratio; FDA=Food and Drug Administration; w=weight; wks=weeks; mon=months; BMI%=BMI percentile.
Active comparator studies
In the active comparator studies cited above, pediatric patients receiving clozapine showed significant increases in mean blood pressure and tachycardia compared with olanzapine (Shaw et al. 2006), but both treatments showed significant increases in triglycerides. In the second treatment refractory schizophrenia study (Kumra et al. 2008a) olanzapine subjects had significantly greater increases in cholesterol levels than those randomized to clozapine.
In the TEOSS trial, there were significant changes in markers of cardiovascular risk with children and adolescents on olanzapine experiencing nearly a 20 mg/dL increase in total cholesterol and nearly 15 mg/dL increase in LDL and a 15 mL/L increase in insulin, whereas levels in youth on risperidone and molindone remained relatively stable (Sikich et al. 2008). Due to a reduced fasting sample, differences in these measures were only significant across the three medication groups, but baseline-to-endpoint changes did not reach significance with olanzapine.
Placebo-controlled studies
In placebo-controlled trials, a marked absence of analyzed or analyzable data was noted (De Hert et al., 2011). Across all trials, olanzapine had the most results available. Triglycerides increased significantly more on olanzapine in comparison to placebo in both placebo-controlled studies (Tohen et al. 2007; Kryzhanovskaya et al. 2009b); glucose and total cholesterol levels increased significantly more compared with placebo in only one study (Tohen et al. 2007). On quetiapine, triglyceride increases were significantly greater compared with placebo in one study (DelBello et al. 2009) and in both Food and Drug Administration registration trials involving quetiapine
Naturalistic cohort studies
The above-mentioned SATIETY study also found significant increases in glucose, insulin and insulin resistance with olanzapine only, in cholesterol (total, LDL and non-high-density lipoprotein) with olanzapine and quetiapine, and increases in triglycerides on olanzapine, quetiapine, and risperidone (Correll et al. 2009). The aripiprazole-treated and the untreated comparison group did not show significant increases in any of the glucose or lipid metabolism parameters, at least during the first 3 months of follow-up. Also, the emergence of metabolic syndrome was low (1.6%), despite rapid and large weight gain in the SGA-treated patients. Longer-term follow-up data in antipsychotic-naïve youth are needed to examine when, in which patients, and on which treatments metabolic syndrome components and the full syndrome begin to develop. In this cohort study, waist circumference, a component of the metabolic syndrome and marker of cardiovascular risk, increased significantly in patients on all medications (Correll et al. 2009). Aripirazole (5.40 [interquartile ranges [IQR]: 2.87–7.93] cm) was roughly similar to quetiapine (5.27 [IQR: 4.07–6.47] cm) and risperidone (5.10 [IQR: 4.49–5.71] cm), whereas olanzapine had the biggest increase (8.55 [IQR: 7.43–9.67] cm). Two important findings emerged from this study (Correll et al. 2009). First, it appears that all antipsychotics can cause relevant weight gain in youth, particularly in the antipsychotic naïve group. Second, although dysregulation in markers of cardiovascular risk, like insulin resistance, cholesterol and triglycerides, seem to generally follow weight gain trajectories, much like in idiopathic obesity, there is also antipsychotic-driven discordance between weight change and metabolic abnormalities. This likely direct, molecular effect seems to aggravate adverse metabolic changes with olanzapine (regarding lipid and glucose metabolism) and quetiapine (regarding lipids) and attenuate (or delay) adverse metabolic effects with rispridone (regarding cholesterol) and aripiprazole (regarding all lipids). However, the longer-term effects require further clarification involving large and generalizable samples, particularly including antipsychotic-naive subjects.
Pharmacoepidemiologic studies
In the pharmacoepidemiologic study cited above, antipsychotic-treated youth had a higher risk of any cardiovascular, cerebrovascular, or hypertensive adverse event if they were on haloperidol (OR: 3.66 [CI: 1.43 to 9.37]) or multiple antipsychotics (OR: 1.72 [CI: 1.10 to 2.69]) (Jerrell and McIntyre 2008). The risk of T2DM was higher in girls (OR: 1.79 [CI: 1.28 to 2.50]), adolescents 13 years or older (OR: 1.52 [CI: 1.08 to 2.13]), and those on multiple antipsychotics (OR: 2.36 [CI: 1.13 to 4.92]). The mean age at initiation of antipsychotic therapy was 11.4±3.4 years; however, the mean age for onset for T2DM was significantly higher, 13.8±3.8 years. Aripiprazole appeared to be protective, with a significant increase in time from the initiation of therapy to the diagnosis of T2DM (hazard ratio [HR] for prolonging the lag time: 35.92; 95% CI: 3.45–373.80) (Jerrell and McIntyre 2008).
The odds of developing dyslipidemia were higher for girls (OR: 2.08 [CI: 1.41–3.03]), adolescents 13 years or older (OR: 2.08 [CI: 1.41–3.03]), and youth on multiple antipsychotics (OR: 5.26 [CI: 1.64–16.82]). The mean age at the initiation of antipsychotic therapy was 11.8±3.6 years, whereas the mean age at onset of dyslipidemia was 14.3±3.9 years. Being on multiple medications conferred a slight increase in risk regarding shorter time to onset of dyslipidemia (HR: 0.44 [CI: 0.22–0.89]). Preexisting obesity and hypertension significantly increased the risk for type 2 diabetes or dyslipidemia (OR: 4.46 [CI: 3.03–6.57]) (Jerrell and McIntyre 2008).
Hypertension was also examined as an outcome with the odds of developing incident hypertension being higher for youth 13 years or older (OR: 2.78 [CI: 1.69–4.55]), but this risk was not related antipsychotic administration. The risk of cardiovascular events, defined as ischemic or pulmonary heart disease, arrhythmias, and cardiomegaly, was increased on haloperidol (OR:4.34 [CI 1.67 to 11.26]) and to a lesser extent on multiple antipsychotics (OR: 1.57 [CI: 0.99 to 2.47]) when entered into the regression model together against congenital heart defects (OR: 3.29 [CI: 2.17–4.99]) substance-related disorders (OR: 1.50 [CI: 1.10–2.04]), as well as other medications, such as mood stabilizers (OR: 1.31 [CI: 1.04–1.65]), tricyclic antidepressants, selective serotonin reuptake inhibitors, and stimulants. When a different regression analysis was run, not including other categories of medication and including preexisting obesity as well as incident metabolic conditions (OR: 2.14 [CI: 1.48–3.10]) and preexisting metabolic conditions (OR: 1.91 [CI: 1.16–3.15]), multiple antipsychotics (OR: 1.65 [CI: 1.05–2.59]), and haloperidol (OR: 3.68 [CI: 1.43–9.45]) were still significant factors (McIntyre and Jerrell 2008). When individual antipsychotics were compared, aripiprazole was associated with the least risk of cardiovascular events, followed by ziprasidone, risperidone, quetiapine, haloperidol, and, finally, multiple medications. Looking at the same data base, Jerrell examined the effects of antipsychotics medications in African-American youth, finding that they had a higher risk of adverse events on antipsychotics (OR: 1.45 [CI: 1.18–1.79]) and that in this population as well multiple antipsychotics increased the chance of developing obesity (OR: 2.29 [CI: 1.43–3.65]) (Jerrell 2010). While these data are useful and the authors show that the adverse metabolic effects are, on average, diagnosed after the administration of antipsychotic medications, the interpretation of these results is complicated by the fact that the “control” individuals did not match on all demographic variables with the antipsychotic treated youth and that the control youth might differ on relevant illness and behavioral factors since they were not clinically started on antipsychotics (Correll 2009). For example, studies have shown associations between weight gain and bipolar disorder (Goldstein et al. 2008) as well as ADHD (Cortese et al. 2008). Further, differences in age and other socioeconomic factors were not well controlled for, calling into question whether the weight gain and metabolic changes reported were due solely to medication or to other factors that have their own relationship with weight gain and are more likely present in individuals on one or multiple antipsychotics. While some of these criticisms can also be leveled at naturalistic studies, the existence of a noncompliant control group that has similar illness characteristics prompting antipsychotic prescribing assures greater comparability between the treated and untreated groups. Moreover, reliance on chart diagnoses to identify cardiometabolic adverse effects is a further problem of pharmacoepidemiologic research.
Studies Targeting the Reduction of Adverse Effects of Antipsychotics on Body Weight and Glucose and Lipid Metabolism in Youth (Behavioral and Pharmacologic)
Randomized controlled studies of adjunctive treatment options for antipsychotic-related weight gain are scant, being almost absent in children and adolescents. In a meta-analysis examining 32 randomized trials (only 2 of which were in children and adolescents, Table 3) published through 2009 covering 15 interventions, only 5 medications outperformed placebo. Among these treatments, metformin (N=7, n=234) was the most effective at reducing weight gain, being associated with 2.94 (CI: 4.89 to 0.99) kg less gain compared with placebo (Maayan et al. 2010a). The other adjunctive treatments that significantly outperformed placebo were d-fenfluramine (N=1, n=16, −2.60 kg [CI: −5.14, −0.06]), sibutramine (N=2, n=55, −2.56 kg [CI: −3.91, −1.22]), topiramate (N=2, n=133, −2.52 kg [CI: −4.87, −0.16]) and reboxetine (N=2, n=79, −1.90 kg [CI: −3.07, −0.72]). The only 2 published, placebo-controlled trials of augmenting agents for antipsychotic-related weight gain in youth tested metformin against placebo and had mixed results (Table 3). One study (Klein et al. 2006) showed metformin to be associated with 4.08±4.06 kg less weight gain than placebo over 16 weeks with a significant difference of 4.65 cm in waist circumference. The other study, however, a prevention study where metformin was started concomitantly with risperidone, showed no separation from placebo (Arman et al. 2008). Although studies of topiramate's primary use in seizures and migraine and for its adjunctive use in psychiatric illness have found it to be tolerated in children (Elterman et al. 1999; Lewis et al. 2009; Wozniak et al. 2009), to date it has not been studied in a randomized placebo controlled trial of antipsychotic weight gain in youth. Studies of topiramate's primary use in seizures and migraine and for its adjunctive use in psychiatric illness have found it to be well tolerated in adults (Ben-Menachem et al. 2008; Cohen et al. under review) and in children (Elterman et al. 1999; Lewis et al. 2009; Wozniak et al. 2009). To date it has not been studied in a randomized placebo-controlled trial of antipsychotic weight gain in youth.
All values represent mean ± standard deviation.
n = total number of subjects randomized in the study.
[nt/np] = number of subjects in the treatment group/number of subjects in the placebo group with the diagnosis.
[n] = number of subjects included in analysis.
Topiramate may hold even more promise as a preventative agent. While the prevention studies with metformin, in which the weight-reducing agent was started concomitantly with the antipsychotic, seemed less effective than the intervention studies, in which weight gain had occurred and, possibly, insulin resistance had developed, the one prevention study with topiramate had the largest effect size in adults (Narula et al. 2010). This suggests that a primary appetite-reducing effect might be superior to an insulin-sensitizing effect for the prevention of antipsychotic-induced weight gain. However, more data are needed to test this hypothesis, and the potential for adverse cognitive effects described for topiramate needs to be assessed and balanced against the potential advantages for reducing weight in patients starting antipsychotic treatment.
In terms of nonpharmacologic strategies to ameliorate antipsychotic-associated weight gain there are no pediatric trials to date. The adult data, however, do suggest considerable efficacy. Ten studies (n=482) (Alvarez-Jimenez et al. 2008), in a recent meta-analysis and systematic review examining behavioral interventions to treat antipsychotic-related weight gain, demonstrated a combined 2.56 kg (CI: 1.92, 3.20) greater reduction in weight compared with treatment as usual. In a subanalysis of individual effect sizes, there was a trend in favor of nutritional counseling (−3.12 kg) over cognitive behavioral therapy (−2.14 kg), without reaching statistical significance. The authors also found, in contrast to data from pharmacologic approaches to reduce weight gain, that there were no significant differences between prevention and intervention trials. In addition, there was no significant difference between individual and group interventions. None of these studies, however, included children (Alvarez-Jimenez et al. 2008).
Discussion
Unfortunately, the pediatric psychiatry literature on the identification and management of cardiometabolic risk factor accrual associated with antipsychotics has many gaps. The most profound limitations appear to be in establishing the different metabolic effects of antipsychotics on drug-naïve youth at various age points. This is important, as age and prior antipsychotic exposure are likely relevant risk factors in the development of weight gain (Correll et al. 2009; Maayan and Correll 2010). Data on the efficacy and safety of switching medications are also needed, as is more research on pharmacologic and nonpharmacologic interventions to ameliorate or, even, prevent weight gain. Further, as in adults receiving antipsychotics, clinical and, especially, biological risk factors for as well as mechanisms of antipsychotic-induced weight gain and metabolic abnormalities require further elucidation (Correll et al. 2011b). On a more basic level, the reporting of weight change and metabolic abnormalities in youth needs to improve. Unfortunately, the vast majority of studies describing antipsychotic-related weight change in youth lack sex- and age-adjusted BMI percentile and BMI z-score measures. While absolute weight change is a valid measure up to approximately 3 months of follow-up, weight change needs to be adjusted to account for physiologically dependent weight change related to normal growth when following pediatric patients over longer periods (Vieweg et al. 2005; Correll and Carlson 2006; Correll 2008). Moreover, in addition to displaying mean changes in weight and in lipid and glucose parameters, shifts into meaningful abnormal categories need to be displayed (Correll 2008).
Current research that will inform some of these issues includes an NIMH-funded study examining the efficacy of risperidone, aripiprazole, and olanzapine in controlling pediatric aggression as well as the effects on clinical function and metabolic side effects that accompany switching to aripiprazole. Metabolic effects are measured using assessments of resting metabolic rate, total body fat, and insulin action in skeletal muscle, liver, and adipose tissue (National Institutes of Health and Washington University School of Medicine 2000).
Regarding clinical interventions to ameliorate weight gain, a multisite, NIMH-funded study is currently underway in children and adolescents with clinically significant weight gain from antipsychotics examining the effects of switching to a possibly less orexigenic antipsychotic (aripiprazole or perphenazine) versus adding metformin versus a control group receiving healthy lifestyle guidance that is provided to all three groups (National Institute of Mental Health 2000). This will be the first study to directly compare a switch to a lower cardiometabolic-risk antipsychotic with adjunctive weight loss medication treatment (National Institute of Mental Health 2000).There is also an ongoing trial to assess an H1 partial agonist, betahistine, with a history of usage in Meniere's disease for pediatric antipsychotic-associated weight gain, based on findings suggesting that H1 antagonism is linked to antipsychotic weight gain (Kroeze et al. 2003; Kim et al. 2007).
Despite these efforts, too little is known regarding the phenomenology and etiology of metabolic effects of antipsychotics in youth. Careful assessments of metabolic parameters of antipsychotics and other psychotropic medications will help to shed light on this important side effect cluster and foster the development of safer effective treatments for child psychiatric disorders.
From a clinical standpoint, proactive, routine monitoring of body weight, BMI percentile and z-score, blood pressure, and fasting glucose and lipids should always be part of the treatment with antipsychotics. Several guidelines and recommendations have been published (e.g., American Diabetes Association et al. 2004; Correll 2008; De Hert et al. 2011). In youth on antipsychotics, weight and height should be measured at each clinical visit and BMI percentile and z-score should be calculated (e.g.,
Conclusions
Because of their efficacy and the prevalence of the disorders they target, pediatric SGA usage has risen dramatically. These medications likely cause age-inappropriate weight gain and disturbances in glucose and lipid metabolisms in all populations, but particularly in young and antipsychotic-naïve individuals. Based on the results from active comparator and placebo-controlled trials as well as cohort studies, it appears that the ranking order of the propensity of these medications to cause clinically significant weight gain in youth is roughly similar to that seen in adults, except that the magnitude of the weight gain is greater and that all antipsychotics seem to have weight gain liability at first antipsychotic exposure. Similar to adult data, clozapine and olanzapine are most orexigenic, followed by risperidone and quetiapine, with aripiprazole and ziprasidone having the lowest risk. Clinical and pharmacoepidemiologic data confirm these findings and suggest that increases in cardiovascular and metabolic risk follow weight gain, with multiple antipsychotics conferring the highest risk. Moreover, some antipsychotics, especially olanzapine and clozapine, seem to have weight independent, direct adverse metabolic effects. Both pharmacologic and behavioral interventions have been little studied in youth, although so far metformin and topiramate seem to have the most promise. Switch studies to a less orexigenic agent along with mechanistic and pharmacologic treatment studies are currently being conducted and will hopefully inform the field to improve the care for youth with psychiatric illness. In addition, cardiometabolic data from newer SGAs approved over the last 2 years also need further study, and lower cardiometabolic risk antipsychotics should be studied head-to-head.
Clinical Significance
Based on the weight and metabolic risk profile of antipsychotics, indications for initial and continued use should be considered very carefully, and proactive cardiometabolic monitoring should be part of the routine clinical treatment with antipsychotics, especially in the more vulnerable pediatric patients. In this context, the documented low monitoring rates in youth are highly concerning (Morrato et al. 2009; 2010a, 2010b), urgently requiring the development of interventions to ameliorate these inadequate monitoring rates in clinical practice. Finally, whenever overweight, obesity, hyperglycemia, dyslipidemia, and hypertension emerge, integrative treatment by mental and physical health care teams needs to be coordinated in order to adequately address these physical health problems with long-term importance.
Footnotes
Disclosures
Dr. Maayan has received research support from Eli Lilly and Pfizer. Dr. Correll has been a consultant to or has received honoraria from Actelion, AstraZeneca, Bristol-Myers Squibb, Cephalon, Eli Lilly, GSK, Intra-Cellular Therapies, Janssen/J&J, Lundbeck, Otsuka, Medicure, Merck, Pfizer, Schering-Plough, Sepracor/Sunovion, Supernus Takeda, and Vanda.
Acknowledgments
This study was supported in parts by The Zucker Hillside Hospital NIMH Advanced Center for Intervention and Services Research for the Study of Schizophrenia MH 074543-01 and by the Stanley Medical Research Institute Award 07TGF-1112. The authors would like to thank Allison Larr for her help in the preparation of this manuscript.