Obesity and schizophrenia: location, location, location?

Obesity is a worldwide epidemic and it is estimated that 300 million people are obese, defined as having a body mass index (BMI) in excess of 30 kg/m² (for review, see Reference Hill, Wyatt and ReedHill et al, 2003). A meta-analysis (Reference Allison, Mentore and MoonseongAllison et al, 1999b ) and review (Reference Taylor and McAskillTaylor & McAskill, 2000) have suggested that antipsychotic drugs – in particular, certain atypical antipsychotic agents – are associated with this weight gain, and schizophrenia has been associated with obesity (Reference Brugha, Wing and SmithBrugha et al, 1989; Reference KendrickKendrick, 1996; Reference Allison, Fontaine and HeoAllison et al, 1999a ). Certain illnesses such as type 2 diabetes mellitus, insulin resistance, dyslipidaemias and cardiovascular disorders, together with obesity, have been termed the metabolic syndrome (Reference ReavenReaven, 1988) and appear to occur more frequently in people with schizophrenia, as has been shown by a recent study conducted in Finland (Reference Heiskanen, Niskanen and LyytikainenHeiskanen et al, 2003). It is believed that obesity-related illnesses may be associated particularly with an increase in visceral fat, the most metabolically active constituent of abdominal obesity (Reference Ryan and ThakoreRyan & Thakore, 2002).

In order to control for the confounding effects of medication, we measured visceral fat distribution using computed tomography in 15 patients with schizophrenia and matched them with healthy controls in terms of age, exercise, diet, smoking habits and alcohol intake (Reference Thakore, Vlahoos and MartinThakore et al, 2002). Seven patients were drug-naïve and the rest had not taken any oral neuroleptic preparation for at least 6 weeks and had had no intramuscular preparation for 6 months; none of the patients had been taking any form of atypical neuroleptic agent prior to entering the study. Patients with schizophrenia had a higher mean BMI than the control group: 26.7 (s.d.=1.1) kg/m² v. 22.8 (s.d.=0.5) kg/m². Patients and controls had similar amounts of total body fat and subcutaneous fat, but the patients had over 3.4 times more intra-abdominal fat than the normal controls: 13 232.0 (s.d.=2666.5) mm² v. 3879.9 (s.d.=571.9) mm² However, there was no difference in intra-abdominal fat distribution between patients who were drug-naïve and those who were drug-free: 12 442.4 (s.d.=9762.6) mm² v. 14 133.9 (s.d.=11 656.8) mm².

An increase in visceral fat is not merely a ‘mass effect’ of a raised BMI; Enzi et al (Reference Enzi, Gasparo and Biondetti1986) found that healthy volunteers with BMI values ≥26 had less intra-abdominal fat (4650 mm²) than the patients in our study (13 232 mm²). Chronically elevated levels of cortisol, also seen in our study, may provide an explanation for the increase in intra-abdominal fat, as the density of glucocorticoid receptors (cytosolic signal transducers for steroids such as cortisol) and the concentrations of the lipogenic enzyme lipoprotein lipase (a key enzyme in fat deposition) are higher in visceral fat than in subcutaneous fat (Reference Ottoson, Vikman-Adolfson and EnerbackOttoson et al, 1994; Reference Pedersen, Jonier and RichelsenPedersen et al, 1994).

Hyperglycaemia, insulin resistance and schizophrenia: an illness effect?

Even though the higher rates of type 2 diabetes mellitus observed in people with schizophrenia have been attributed to the use of antipsychotic medications – in particular, atypical agents – this is by no means a universally accepted finding. For instance, Mukherjee et al (Reference Mukherjee, Decina and Bocola1996) studied a cohort of patients with schizophrenia (n=95), and observed that the prevalence of diabetes was age-dependent and greater in those taking conventional neuroleptic medications. Subramaniam et al (Reference Subramaniam, Chong and Pek2003) reported a rate of undiagnosed diabetes mellitus of 16% and a rate of impaired glucose tolerance of just over 30% in a cohort of residential patients with schizophrenia, none of whom had ever received an atypical neuroleptic drug; yet the rate of type 2 diabetes mellitus in the general population of a similar age was over 22%, indicating that patients with schizophrenia are less likely to have their diabetes diagnosed than their counterparts without mental illness.

The introduction of atypical neuroleptics has added to this debate, although most of the evidence implicating these compounds is based on case reports and various cross-sectional epidemiological studies (Reference Liebzeit, Markowitz and CaleyLiebzeit et al, 2001; Reference Sernyak, Leslie and AlarconSernyak et al, 2002). In contrast to these findings, Lieberman et al (Reference Lieberman, Phillips and Hongbin2003) conducted a prospective study in a Chinese population, comparing chlorpromazine with clozapine in drug-naïve patients with first-episode schizophrenia over a 52-week period, and showed that despite significant increases in weight (which were equal between the two compounds in question), there was no significant increase in fasting plasma glucose levels at the end of the study period. However, the study did not have a normal control group as a reference population. This is important, because the rates of obesity and type 2 diabetes mellitus in this population are lower than those found in North America, or indeed in Europe. Furthermore, lifestyle issues such as diet and exercise were not discussed either before or during the treatment period.

Is it possible that a mechanism other than medication might be responsible for such findings? A number of papers from the era before the use of antipsychotic drugs add credence to this hypothesis, although problems with diagnosis, small size of study group and other methodological issues make it difficult to interpret the significance of these valuable earlier studies (Reference LorenzLorenz, 1922; Reference Braceland, Meduna and VaichulisBraceland et al, 1945; Reference FreemanFreeman, 1946; Reference LangfeldtLangfeldt, 1952). It is notable that a family study found that up to 19% of first-degree relatives of patients with schizophrenia had type 2 diabetes mellitus, which indicates that this endocrine condition and schizophrenia might have a genetic association (Reference Mukherjee, Schnur and ReddyMukherjee et al, 1989).

In an attempt to determine whether schizophrenia is associated with abnormal glucose metabolism, we compared fasting levels of plasma glucose, insulin, lipids and cortisol measures in a group of hospitalised, drug-naïve patients with first-episode schizophrenia (n=26) with those of a healthy volunteer group matched in terms of age, ethnicity, exercise, diet, smoking habits and alcohol intake (Reference Ryan, Collins and ThakoreRyan et al, 2003). Anthropometric and lifestyle data indicated that the only significant difference between the two groups was that patients had a higher saturated fat intake than did controls. Over 15% of patients with schizophrenia had impaired fasting glucose levels – compared with none in the control group – as defined by the American Diabetes Association (1997) criteria. Patients with schizophrenia, compared with the control group, had significantly higher plasma levels of fasting glucose (5.3 (s.d.=0.9) mmol/l v. 4.8 (s.d.=0.3) mmo/l), insulin (68.2 (s.d.= 64.6) pmol/l v. 55.2 (s.d.=26.5) pmol/l) and cortisol (499.4 (s.d.=161.4) nmol/l v. 303.2 (s.d.=10.5) nmol/l), and were more insulin-resistant: 2.3 (s.d.=1.0) v. 1.7 (s.d.=0.7). Both the control and the patient groups had similar levels of lipids. Finally, there was no significant association between severity of symptoms and plasma levels of glucose, indicating that the ‘stress of hospitalisation’ was an unlikely cause of the hyperglycaemia.

The rate of impaired fasting glucose concentration observed in our group of patients (>15%) is greater than that found in a recent European study (8.5%, Reference Gourdy, Ruidavets and FerieresGourdy et al, 2001). Type 2 diabetes mellitus and vascular complications occur in a third of those with impaired fasting glucose levels (Reference AlbertiAlberti, 1996). Medication, age, ethnicity, physical inactivity and smoking are unlikely to explain our findings (Reference KingKing & WHO Ad Hoc Reporting Group, 1993; Reference Shaten, Kuller and Davey SmithShaten et al, 1993). Although our patients consumed more saturated fat, studies do not indicate a positive association between a high intake of saturated fat and hyperglycaemia (Reference Colditz, Manson and StampferColditz et al, 1992; Salmeron et al, Reference Salmeron, Ascherio and Rimm1997, Reference Salmeron, Hu and Manson2001), however, patients with schizophrenia did have higher levels of cortisol than did normal controls.

Are patients with schizophrenia biologically stressed?

A common endocrine reaction to stress involves activation of the hypothalamic–pituitary–adrenal (HPA) axis (Reference Axelrod and ReisineAxelrod & Reisine, 1984). As in Cushing's syndrome and melancholic depression (Reference Wajchenberg, Bosco and MaroneWajchenberg et al, 1995; Reference Condren and ThakoreCondren & Thakore, 2001; Reference Thakore, Vlahoos and MartinThakore et al, 2002), a dysregulated HPA axis can lead to abnormal glucose metabolism and visceral obesity (Reference Rosmond and BjorntorpRosmond & Bjorntorp, 2002). Schizophrenia is associated with abnormalities of this axis (Reference Altamura, Guercetti and PercudaniAltamura et al, 1989; Reference Coryell and TsuangCoryell & Tsuang, 1992; Reference Kaneko, Yokoyama and HoshinoKaneko et al, 1992; Reference Lammers, Garcia Boreguero and SchmiderLammers et al, 1995), and we have confirmed this using a rather crude indicator of HPA axis activity in two studies (Reference Thakore, Vlahoos and MartinThakore et al, 2002; Reference Ryan, Collins and ThakoreRyan et al, 2003).

To date, HPA axis disturbance has been less consistently reported in schizophrenia than in depression (Reference Holsboer and GrossmanHolsboer, 1998; Reference Cotter and ParianteCotter & Pariante, 2002). With respect to schizophrenia, adrenocorticotrophic hormone (ACTH) and cortisol responses to corticotrophin-releasing hormone (CRH) are indistinguishable from controls, although pre-treatment with dexamethasone results in an exaggerated CRH-induced pituitary–adrenal response in patients (Reference Roy, Pickar and DoranRoy et al, 1986; Reference Lammers, Garcia Boreguero and SchmiderLammers et al, 1995). Most (but not all) studies have shown that dexamethasone suppresses plasma levels of cortisol in patients with schizophrenia (Reference Dewan, Pandurangi and BoucherDewan et al, 1982; Reference Tandon, Mazzara and De QuardoTandon et al, 1991). Equally discordant findings have been reported in terms of basal activity of the HPA axis as measured by serum cortisol levels (Reference Gil-Ad, Dickerman and AmduskyGil-Ad et al, 1986; Reference Roy, Pickar and DoranRoy et al, 1986; Reference Whalley, Christie and BlackwoodWhalley et al, 1989; Reference Van Cauter, Linkowski and KerkhofsVan Cauter et al, 1991; Reference Breier and BuchananBreier & Buchanan, 1992; Reference Rao, Strebel and HalarisRao et al, 1995; Reference Elman, Adler and MalhotraElman et al, 1998; Reference Kaneda, Fujii and OhmoriKaneda et al, 2002). Methodological problems may partly explain the differences observed between the studies quoted. For instance, the effects of medication on HPA axis activity are unclear (Reference HellewellHellewell, 1999), and often a single sample of cortisol has been used to determine HPA activity although it is not clear whether this accurately represents an estimate of mean 24 h activity (Reference Muller, von Werder, Muller and MacleodMuller & von Werder, 1989).

As mean or integrated measures, such as area under the curve (AUC), of plasma cortisol between 13.00 h and 16.00 h can be used to detect hypercortisolism (Reference Halbreich, Zumoff and KreamHalbreich et al, 1982), we decided to determine cortisol, ACTH and arginine vasopressin (AVP) levels in drug-naïve patients with first-episode schizophrenia and compare them with a group of volunteers matched for age and gender (Reference Ryan, Sharifi and CondrenRyan et al, 2004). Baseline levels of cortisol and AVP were indistinguishable between patients and controls, although patients had higher ACTH levels. Patients with schizophrenia had a higher mean AUC of ACTH (26.3 (s.d.=6.2) nmol/l v. 13.9 (s.d.=3.0) nmol/l) and cortisol (279.4 (s.d.=26.0) nmol/l v. 213.1 (s.d.=18.4) nmol/l) but had a lower mean AUC of AVP (0.87 (s.d.=0.24) pmol/l v. 1.42 (s.d.=0.34) pmol/l) than controls. A positive correlation between plasma levels of AVP and cortisol, and higher levels of plasma ACTH during the test period, indicate that the pituitary–adrenal axis was more sensitive to vasopressin-mediated stimulation in our patients with schizophrenia. This may be due first to the fact that vasopressin can directly stimulate the release of cortisol from the adrenal cortex (Reference Guillon, Trueba and JoubertGuillon et al, 1995), and second, to the fact that glucocorticoid-induced inhibition of AVP gene transcription may be overcome, thereby allowing this hypothalamic neuropeptide to stimulate the pituitary–adrenal axis (Reference Rivier and ValeRivier & Vale, 1983; Reference Kovacs and SawchenkoKovacs & Sawchenko, 1996; Reference Aguilera and Rabadan-DiehlAguilera & Rabadan-Diehl, 2000; Reference Aguilera, Lightman, Ma, McCarty, Aguilera and KvetnanskyAguilera et al, 2000), leading to a relative hypercortisolaemia with all its consequent effects.

CLINICAL IMPLICATIONS

1. ▪ Drug-naïve patients with first-episode schizophrenia may have important metabolic disturbances, including central obesity and impaired fasting glucose levels.

2. ▪ Clinicians should be aware of the cardiovascular complications associated with such metabolic disturbances and ensure that their patients have regular contact with their general practitioner – or indeed a diabetologist.

3. ▪ Appropriate clinicians should not only monitor plasma glucose levels but also check for signs of central obesity by measuring waist–hip ratios at diagnosis and also at regular intervals thereafter.

LIMITATIONS

1. ▪ The definitions of diabetes and schizophrenia used by earlier researchers would not conform to the rigour of modern standards, and therefore the observations and rates quoted may not be wholly accurate.

2. ▪ The numbers of patients used in the studies described were small and it may be difficult to extrapolate these findings to larger populations. Larger prospective studies need to be performed.

3. ▪ A literature search over such a broad area cannot be regarded as fully comprehensive. Some papers were not translated from their original language.