- Email: [email protected]
Altered glucose tolerance in women with deliberate self-harm
Psychoneuroendocrinology (2009) 34, 878—883
a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m
j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / p s y n e u e n
Altered glucose tolerance in women with deliberate self-harm ´n b, Charlotta Sunnqvist a, Lil Tra Sofie Westling a,*, Bo Ahre ¨skman-Bendz a a b
Department of Clinical Sciences, Psychiatry, Lund University, 221 85 Lund, Sweden Department of Clinical Sciences, Medicine, BMC, Lund University, 221 84 Lund, Sweden
Received 23 September 2008; received in revised form 1 December 2008; accepted 30 December 2008
KEYWORDS Glucose tolerance test; Deliberate self-harm; Violent; Insulin; Glucagon; Borderline personality disorder
Summary Disturbances in glucose metabolism are of importance for violent behaviour in men, but studies in women are lacking. We used the 5 h-oral glucose tolerance test (OGTT) in this study of 17 female psychiatric patients, selected for violent behaviour directed against themselves (deliberate self-harm) and 17 healthy controls matched for age and BMI. Following OGTT, patients had higher glucose levels at 30 min ( p = 0.007) and increased glucagon area under the curve ( p = 0.011). Since a co-morbid eating disorder might affect results, we as a post-hoc analysis subgrouped the patients and found that the increased glucagon levels only were present in patients with an eating disorder. In contrast, those without an eating disorder showed a significantly lower p-glucose nadir ( p = 0.015) and unaltered glucagon levels compared to controls. There were no significant differences in insulin and C-peptide levels between patients and controls. We conclude that deliberate self-harm in women may be associated with alterations in carbohydrate metabolism in certain groups. Eating disorder is a confounding factor. # 2009 Elsevier Ltd. All rights reserved.
1. Introduction Several studies have investigated the relationship between glucose metabolism and impulsive aggressive behaviour in healthy men as well as in male forensic psychiatry patients. During the oral glucose tolerance test (OGTT), a reactive hypoglycaemic tendency after the initial rise in plasma (p)glucose has been found in habitually violent men (Bolton, 1973; Benton et al., 1982; Virkkunen and Huttunen, 1982; Virkkunen, 1982, 1983, 1986; Virkkunen and Narvanen, 1987; Virkkunen et al., 1994). In some of these men hypo-
* Corresponding author. Tel.: +46 735 62 60 99; fax: +46 46 17 38 44. E-mail address: [email protected] (S. Westling).
glycaemia has persisted throughout the OGTT. The counterregulation of hypoglycaemia is a complex system, involving a decrease in insulin secretion and an increase in several counter-regulatory factors, augmenting p-glucose, of which the hormone glucagon is the most potent during the immediate phase (Cryer, 1993). The hypoglycaemic tendency found in habitually violent men may thus be explained by either an imbalanced insulin secretion, which does not decrease when p-glucose is reduced, or an insufficient secretion of the counter-regulatory factors raising pglucose. Early studies by Virkkunen and coworkers have found increased insulin levels, proposing this as the cause of the hypoglycaemia (Virkkunen, 1982, 1983). A recent study, examining men with habitually violent behaviour, found that they, compared with healthy controls, had
0306-4530/$ — see front matter # 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.psyneuen.2008.12.015
Altered glucose tolerance in women with deliberate self-harm significantly lower basal glucagon levels, which could be a possible explanation for their hypoglycaemic tendencies (Virkkunen et al., 2007). The men in these studies often showed self-aggression and had made several violent suicide attempts (Virkkunen and Narvanen, 1987). In psychiatric care, violent behaviour among patients is often directed towards the self, and includes deliberate self-harm as well as attempted suicide. This kind of violent behaviour is commonly, but not exclusively, associated with borderline personality disorder, a comorbid diagnose in many of the patients studied by Virkkunen (1986). Furthermore, epidemiological studies among adolescents in the United States have found that aggressive behaviours directed against oneself, and aggressive behaviours towards other people often occur in the same individuals (Borowsky et al., 2001; Centers for Disease Control and Prevention, 2004). Treatment of deliberate self-harm is traditionally psychotherapeutic although hospitalization as well as polypharmacy combining anti-depressant, anti-psychotic as well as tranquilizing medication is common. In patients with severe deliberate self-harm, treatment effect is often limited, and mortality rates in completed suicide are high. Examining glucose metabolism in these patients is a completely new approach and if altered glucose metabolism is found, new treatments might develop as complement to the present ones. Whereas disturbed glucose metabolism has been found in aggressive men, it is not known whether this is the case also in women. We therefore investigated insulin and glucagon secretion and glucose tolerance during OGTT in women with deliberate self-harm. Our primary hypotheses were: (1) Women with deliberate self-harm have high self-rated physical aggression. (2) Women with deliberate self-harm have lower p-glucose nadir (lowest value measured during the OGTT) compared to healthy controls. (3) Women with deliberate self-harm have increased insulin secretion during OGTT compared to healthy controls. (4) Women with deliberate self-harm have decreased glucagon secretion compared to healthy controls.
2. Methods 2.1. Participants From February 2005 to December 2007, female patients with current deliberate self-harm, attending the Division of Psychiatry at the University Hospital in Lund were asked to participate in the study. Inclusion criteria were a history of deliberate self-harm (Hawton et al., 2002) for at least 2 years with at least five incidents occurring during the last 6 months. In order to try to make the group more homogeneous we decided only to include patients with a borderline personality disorder, a condition associated with violent behaviour directed towards self or others. Exclusion criteria were diabetes mellitus or an active liver disease (defined as a twofold rise in p-alanine aminotransferase (ALAT) or p-aspartate aminotransferase (ASAT). Seventeen patients were finally included. Controls matched for sex, age (8 years) and body mass index
879 (BMI) (2 kg m 2) were recruited from a random selection of the population register in the city of Lund and non-randomly among students at the University of Lund. The patients were diagnosed according to the Diagnostic and Statistical Manual of Mental Disorders, fourth edition, axis I, II and III (American Psychiatric Association, 2000). All subjects were classified and rated by the corresponding author. All patients had a borderline personality disorder according to the Structured Clinical Interview for DSM IVAxis II disorders (First et al., 1998). Two patients fulfilled the criteria for one axis I diagnosis, six fulfilled the criteria for two, further six fulfilled the criteria for three, and three patients fulfilled the criteria for four axis I diagnoses. Twelve of the patients had anxiety disorders, 10 had a mood disorder, seven had an eating disorder, seven had a substance use disorder (four with alcohol abuse), two had an Attention Deficit Hyperactivity Disorder, one had a dissociative disorder, and one had psychotic symptoms not other specified. Thirteen patients had somatic diagnoses, of which asthma/ allergy (n = 6) and hypothyroidism (n = 2) were the most common ones. All patients used medication. Fifteen used a psychotropic medication; no patient used lithium. The most common medication for somatic diseases were inhalators/ anti-allergic medication (n = 6), contraceptives (n = 5) and omeprazol (n = 4). Six of the healthy controls used hormonal contraceptives and five used nutritional supplements. The patients’ medications had not been changed for a median of 33 days (interquartile range 24—88 days). Only nine of the 17 patients had regular menstruation.
2.2. The Aggression-Questionnaire Revised Swedish Version (AQ-RSV) Level of aggression was estimated by use of the AQ-RSV (Prochazka and Agren, 2001). AQ-RSV consists of 29 items, distributed into four scales, measuring the aggression factors physical aggression (nine items), verbal aggression (five items), anger (seven items) and hostility (eight items). Each item is rated on a 4-point scale, from the least to the most characteristic of the subject. We used the scale for Physical aggression in order to see if our subjects could be regarded as comparable to the habitually violent men in the studies by Virkkunen and co-workers. The raw score presented in the table is the total sum of the ratings of items included in the scale for physical aggression. In the English version of the aggression questionnaire, the sub-scale physical aggression has a reliability of 0.80 and significant correlations have been found between the sub-scale and peer nomination (Buss and Perry, 1992). Reliability and validity has not been tested in the Swedish version. However, the internal consistency of the four aggression subscales and the total score was evaluated by the alpha coefficients (Cronbach’s alpha). The alphas of physical aggression were comparable in the American and Swedish populations and the alpha coefficients indicated considerable internal consistency. Normative data for AQRSV has been obtained earlier from the general population (Prochazka and Agren, 2001). We used a formula developed by Helena Prochazka (personal communication) to compare our patients to women in the general Swedish population. AQ-RSV has not previously been used on patients with deliberate self-harm.
880
S. Westling et al.
2.3. Oral glucose tolerance test and sampling The subjects arrived after one night of fasting. Blood samples were examined for possible infections, anaemia, hepatic and thyroid diseases, plasma-lipids and electrolytes. Test samples included glucose, insulin, c-peptide and glucagon. After baseline sampling the subjects ingested 300 ml solution-containing water mixed with 75 g glucose. New plasma samples were drawn after 15, 30, 45, 60, 75, 90, 120, 150, 180, 240 and 300 min. Subjects were asked to report any adverse symptoms and additional samples for immediate analysis of capillary blood glucose were drawn. The test samples were collected in tubes containing EDTA (7.4 mmol/l; final concentration) and aprotinin (500 kallikrein inhibitor units/ ml blood; Novo Nordisk, Bagsvaerd, Denmark). 10,000 IE trasylol/ml blood was added to the tube collected for glucagon analysis. All test samples were immediately put on ice and centrifuged at +4 8C, 3000 rpm, within at most 3 h. Plasma was frozen at 80 8C until analysis. Insulin, C-peptide and glucagon concentrations were analyzed with double antibody radioimmunoassay techniques (Linco Research, St. Charles, MO); glucose was determined with the glucose oxidase method.
2.4. Statistical analyses A power analysis was made before the study was initiated. We used p-glucose nadir as a primary effect variable, which in former studies have shown a standard deviation of 0.5. With a clinically relevant difference of 0.5 mM and power of 0.8, a significance level of 5% demands a sample of 17 matched couples. The software SPSS 13.0 for Windows was Table 1
used for all statistical calculations. Because of the small sample size we used non-parametric statistics. p-Values were calculated with Mann—Whitney U-test. Results are presented as median and interquartile range. BMI was calculated by dividing the patients’ weight in kg by the squared height. Area under curve (AUC) was calculated with the trapezoid method. As a measure of insulin sensitivity, we used insulinogenic index defined as (insulin30 min insulin0 min) (glucose30 min glucose0 min) 1 (Ahren et al., 2008). One patient had a missing value for glucagon and one had a missing value for c-peptide during the glucose tolerance test. When calculating the AUC for glucagon and c-peptide, respectively in these patients, the missing values were estimated to be the mean of the values immediately preceding and following the missing one. One patient had missing values for the last observation (300 min). This patient only completed 240 min of the protocol due to severe suicidal impulses, increasing during the test. The values for the sample at 300 min were then calculated by dividing the remaining 33 subjects’ values for 300 min by the values for 240 min and then multiplying the quotient with the patient’s observed values at 240 min. The quotients were all normally distributed, within a reasonable range. Two patients had each one missing value on an item in the AQ-RSV. The score of this item was then calculated as the median of the scores from the remaining eight items.
2.5. Ethical approval The study was approved by the Regional Ethical Review Board in Lund and all subjects gave written informed consent before the study.
Comparisons between patients and healthy controls (median and interquartile range).
Age (years) BMI (kg m 2) p-GT (glutamyltransferase) (ref < 0.4 mkat/l) p-ALAT (alanine aminotransferase) (ref < 0.7 mkat/l) fp-Glucose (mM) p-Glucose 30 min (mM) p-glucose 300 min (mM) p-Glucose nadir (mM) p-Glucose AUC (M min) fp-Insulin (pM) p-Insulin 30 min (pM) p-Insulin AUC (mM min) Insulinogenic index 30 min (pmol mmol 1) a fp-c-Peptide (nM) p-c-Peptide 30 min (nM) p-c-Peptide AUC (nM min) fp-Glucagon (ng l 1) p-Glucagon 30 min (ng l 1) p-Glucagon AUC (mg l 1) AQ-RSV, physical aggression score
Patients (n = 17)
Controls (n = 17)
p
20 (19—28) 22.1 (18.8—25.8) 0.26 (0.20—0.32) 0.25 (0.22—0.37) 4.7 (4.1—5.2) 8.2 (7.2—9.2) 4.2 (3.7—4.6) 3.2 (2.8—3.6) 1.6 (1.4—1.7) 70 (43—90) 526 (361—738) 67 (39—100) 106 (81—180) 0.41 (0.34—0.86) 2.14 (1.44—3.26) 451 (276—586) 69 (46—86) 60 (50—78) 16 (15—23) 23 (16—28)
24 (21—27) 21.1 (19.1—25.2) 0.19 (0.16—0.26) 0.25 (0.20—0.38) 4.6 (4.2—5.1) 7.0 (5.6—7.6) 4.6 (4.0—4.8) 3.9 (2.7—4.2) 1.5 (1.3—1.7) 80 (61—85) 471 (301—826) 67 (45—86) 207 (99—327) 0.54 (0.40—0.75) 2.35 (1.52—3.5) 456 (321—528) 46 (37—66) 46 (37—65) 13 (10—15) 11 (9—12)
0.218 0.892 0.110 0.817 0.946 0.007 ** 0.204 0.231 0.160 0.432 0.812 0.973 0.377 0.766 0.786 0.865 0.073 0.053 0.011 * <0.001
Abbreviations: BMI: body mass index; fp: fasting plasma; p: plasma; AQ-RSV: Aggression Questionnaire Revised Swedish Version. a Insulinogenic index = (insulin30 min insulin0 min) (glucose30 min glucose0 min) 1 (Ahren et al., 2008). * p = 0.011, Mann—Whitney U-test. ** p = 0.007, Mann—Whitney U-test.
Altered glucose tolerance in women with deliberate self-harm
3. Results 3.1. Patients (n = 17) and controls (n = 17) (Table 1) 3.1.1. Basal characteristics There were no significant differences between patients and controls in BMI, age, p-glutamyltransferase (GT) or p-ALAT. 3.1.2. AQ-RSV Patients had significantly higher scores on physical aggression (23 [16—28]) compared to controls (11 [9—12]) ( p < 0.001). When relating to women in the general Swedish population, our patients had higher scores with a ratio of 1.45:1 (1.04— 1.74:1). 3.1.3. Glucose There was no significant difference between patients and controls in fasting plasma (fp-) glucose. Following the glucose administration, glucose levels increased to a peak after 30 (30—45) min; and thereafter glucose levels fell. After 180 (120—180) min, a nadir was reached. Patients had significantly higher p-glucose after 30 min (8.2 [7.2—9.2] mM) compared to controls (7.0 [5.6—7.6] mM) ( p = 0.007). There were no significant differences in p-glucose nadir, p-glucose at 300 min or in glucose AUC between patients and controls. There was no significant correlation between p-glucose nadir and scores on physical aggression. 3.1.4. Insulin There were no significant differences in fp-insulin, p-insulin 30 min or insulin AUC between patients and controls. No significant difference was found between patients and controls in insulin sensitivity measured as insulinogenic index. 3.1.5. C-peptide There were no significant differences between patients and controls in fp-c-peptide, p-c-peptide 30 min or c-peptide AUC. 3.1.6. Glucagon There were no significant differences between patients and controls in fp-glucagon or p-glucagon 30 min although a trend could be seen towards increased 30 min levels in patients (60 [50—78] ng l 1) compared to controls (46 [37—65] ng l 1) ( p = 0.053). Patients had a significantly increased glucagon AUC (16 [15—23] mg min/l) compared to controls (13 [10— 15] mg min/l) ( p = 0.011).
3.2. Patients without an eating disorder (n = 10) and controls (n = 10) Patients without an eating disorder showed significantly increased scores on AQ-RSV Physical aggression (24 [20— 29]) compared to controls (11 [9—12]) ( p = 0.001). These patients had significantly lower p-glucose nadir (3.2 [2.8— 3.6] mM) than controls (4.0 [3.4—4.2] mM) ( p = 0.015). They also had lower p-glucose at the endpoint (300 min) (4.0 [3.3— 4.6] mM) than controls (4.7 [4.3—5.2] mM) ( p = 0.029). There was no significant difference between patients and controls in p-glucose 30 min or glucagon AUC. None of the other
881 parameters measured showed any significant differences between patients and controls. There was no significant correlation between p-glucose nadir and scores on physical aggression.
3.3. Patients with an eating disorder (n = 7) and controls (n = 7) Patients with an eating disorder showed significantly increased scores on AQ-RSV physical aggression (22 [14— 24]) compared to controls (11 [9—12]) ( p = 0.012). These patients showed significantly increased glucagon AUC (17 [16—29] mg min/l) compared to controls (13 [12— 21] mg min/l) ( p = 0.038). They also showed a trend towards increased p-glucose 30 min (7.8 [7.2—9.4] mM) compared to controls (7.0 [5.0—7.4] mM) ( p = 0.053). None of the other parameters measured showed any significant differences between patients and controls.
4. Discussion We have investigated glucose, insulin, c-peptide and glucagon in plasma during OGTT and found increased glucagon levels and increased p-glucose after 30 min in women with deliberate self-harm compared to healthy controls. We also investigated all subjects with a self-rating scale for physical aggression and found significantly increased scores in patients. Although no study has investigated a correlation between observed violent behaviour and high ratings on physical aggression on the AQ-RSV, the high scores obtained by our patients is an indicator that this sample might be somewhat comparable to the habitually violent men studied by Virkkunen and co-workers (Virkkunen and Huttunen, 1982; Virkkunen, 1982, 1983, 1986; Virkkunen and Narvanen, 1987; Virkkunen et al., 1994). After the initial calculations, and due to heterogeneity of the patient population, we performed post hoc analyses to identify factors, which could explain differences in subgroups. We thus analysed individual curves for the patients and controls, as well as at groups of curves including patients with specific diagnoses, which we thought might confound our results (i.e. eating disorders, MDD, substance abuse). We found that patients with an eating disorder had elevated glucagon curves. This was not observed in other groups. As a post hoc analysis we therefore separately analysed the results from patients with versus without an eating disorder. Our finding of an increased p-glucagon AUC was confirmed only in patients with an eating disorder. These patients further showed a trend towards increased p-glucose at 30 min. We consider this trend to be subsequent to the rise in glucagon secretion, since the whole glucose AUC seemed elevated in these patients compared to controls, however not significantly. Among patients without an eating disorder, we found a low p-glucose nadir when compared with controls. Our primary hypothesis was therefore confirmed, but only in this subgroup of patients. In addition, patients not suffering from an eating disorder had lower p-glucose than controls at endpoint during OGTT. Thus, regardless of co-morbidity and concomitant medication, women with deliberate self-harm without an eating disorder showed changes in glucose metabolism, com-
882 parable to those found in men with habitually violent behaviour (Virkkunen and Huttunen, 1982; Virkkunen, 1982, 1983, 1986; Virkkunen and Narvanen, 1987; Virkkunen et al., 1994). Aggression is one out of several symptoms of hypoglycaemia (McAulay et al., 2001). Our patients without an eating disorder had a p-glucose nadir of 3.2 mM, which is significantly lower compared to 4.0 mM in controls. This difference is of approximately the same magnitude as earlier found in men, but absolute glucose levels are higher in our patients as well as in our controls (Virkkunen and Huttunen, 1982; Virkkunen, 1982, 1986; Virkkunen et al., 1994). The cause of this difference between absolute values in our study and the Virkkunen ones is not known but is intriguing since not only the patients, but also controls showed higher levels in our study. We did not find any significant correlation between glucose nadir and the ratings for Physical aggression. A causal relation between the lower p-glucose and the aggressive behaviour seem less likely and a common mechanism behind both cannot be excluded. Epinephrine, growth hormone as well as cortisol are factors involved in glucose counterregulation (Cryer, 1993) but not measured in this study. Low pre- and postdexamethasone cortisol has been found in suicide attempters with cluster B personality disorder compared to suicide attempters and healthy controls (Westrin et al., 2003). Furthermore cortisol as well as epinephrine participates in mood regulation. Alterations in these substances might explain the difference found in p-glucose nadir between our patients and controls. However, glucagon is known to be the most potent counter-regulating factor, which means that a normal secretion of glucagon should have prevented hypoglycaemia (Cryer, 1993). Virkkunen et al. (1995) have proposed serotonin as a common denominator, as it might affect behaviour as well as glucose homeostasis. Low concentrations of the serotonin metabolite 5-hydroxyindole acetic acid in cerebrospinal fluid has furthermore been found in impulsive violent men prone to hypoglycaemia (Virkkunen et al., 1994). These are thus substances of interest for future research. Our hypothesis of enhanced insulin secretion in patients could not be verified. Nor did we find any significant differences in insulin sensitivity, explaining the low p-glucose nadir. In studies investigating male forensic patients, results have been somewhat divergent. Enhanced p-insulin during OGTT was found in offenders with intermittent explosive disorder, but not in men with violent antisocial personality disorder (Virkkunen, 1986). The studies on male offenders differ from ours, since the offenders were imprisoned during 5 months before examination and thus abstinent from alcohol and drugs and with a reasonably controlled intake of food and psychotropic drugs. Another difference is that in our study, but not in the studies by Virkkunen et al., not only p-insulin but also c-peptide was measured. C-peptide, in contrast to insulin, is not subjected to a substantial uptake by the liver and thus more relevant when estimating insulin secretion. However, since our study lacked control over a number of mechanisms affecting glucose metabolism, a true difference in insulin sensitivity cannot be excluded. Another important finding in this study is that patients with an eating disorder had an increased glucagon secretion and a trend towards increased p-glucose 30 min, in contrast to patients without an eating disorder. The elevated glucagon
S. Westling et al. secretion would serve to raise p-glucose, thus explaining the differences in p-glucose 30 min as well as in p-glucose nadir between the two subgroups. Eating disorder per se is however a less likely explanation for the glucagon secretion (Mitchell and Bantle, 1983; Kumai et al., 1988; Casper, 1996; Russell et al., 1996; Tomasik et al., 2005; Yasuhara et al., 2005). The most resonable explanation for the elevated glucagon secretion is the lack of control over dietary intake in our study. The irregular and inadequate meals compatible with an eating disorder probably affects and confounds these results. Our patient group was homogeneous for deliberate selfharm and borderline personality disorder, but heterogeneous for a number of factors including concomitant diagnoses, medication, intake of illegal drugs and alcohol, diet and menstruation. This is an inevitable consequence of the study subject; deliberate self-harm, which is a cross-diagnostic behaviour of major clinical relevance. Concerning co-morbidity, our participants well represent patients with deliberate self-harm and borderline personality disorder in a naturalistic setting (Philipsen et al., 2008). Our study thus approximates real life better than it would have done with more stringent exclusion criteria. Another limitation is that all patients, but one, were outpatients. This makes the circumstances during the days preceding the test less controllable. Concurrent medication was prevalent but relatively stable. Some anti-psychotic as well as anti-depressant medication can induce obesity, with subsequent insulin resistance and type 2 diabetes mellitus (T2DM) (Newcomer, 2004; Derijks et al., 2008). Oral contraceptives are also related to an increased risk of impaired glucose tolerance (Pandit et al., 1993). However, we matched controls for BMI and none of the subjects showed either insulin resistance or T2DM. The co-occurrence of diseases other than eating disorder that might affect glucose metabolism is a further limitation. Major depressive disorder (MDD) is associated with insulin resistance (Wright et al., 1978; Winokur et al., 1988). Six of our patients suffered from MDD but none of them showed insulin resistance. Four of the patients were diagnosed with alcohol abuse. However, no participants showed increased levels of p-carbohydrate deficient transferrin (CDT) and active liver disease (as defined in methods) was an exclusion criterion. We have thus tried to minimize the effect of a comorbid alcohol abuse. Another limitation is the influence of the menstrual cycle on glucose tolerance (Escalante Pulido and Alpizar Salazar, 1999). Since only 9 of the 17 patients had regular menstruation we could not to co-ordinate the patients’ menstrual cycles with those of the controls. In conclusion our results seemingly confirm and extend findings by others of alterations in carbohydrate metabolism in persons with violent behaviour, here mainly shown as deliberate self-harm. We suggest that impulsive aggression in female patients is associated with a deviant carbohydrate metabolism with impaired glucagon response to reactive glucose nadir after oral glucose challenge and that a combined eating disorder restores this by enhancing glucagon secretion.
Role of funding source The work was supported by grants from the Swedish Research Council no. 14548 and 6834, Lund University Hospital Research Funds, the Faculty of Medicine, Lund University,
Altered glucose tolerance in women with deliberate self-harm the So ¨derstro ¨m-Ko ¨nigska Foundation, the Sjo ¨bring Foundation, the OM Persson Foundation, the Lundbeck Foundation, Swedish Psychiatry Foundation and Region Ska ˚ne. None of the funding organisations had any further role in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication.
Conflict of interest All authors declare that they have no conflicts of interest.
Acknowledgements The authors are grateful to Lilian Bengtsson and Kristina Andersson for expert technical assistance in the analyses of samples, to Helena Prochazka for supplying profound knowledge on the AQ-RSV and to Hans Bendz for important comments on the manuscript.
References Ahren, B., Pratley, R.E., Soubt, M., Dunning, B.E., Foley, J.E., 2008. Clinical measures of islet function: usefulness to characterize defects in diabetes. Curr. Diab. Rev. 4, 129—145. American Psychiatric Association, 2000. Diagnostic and Statistical Manual of Mental Disorders fourth edition. American Psychiatric Press, Washington, DC. Benton, D., Kumari, N., Brain, P.F., 1982. Mild hypoglycaemia and questionnaire measures of aggression. Biol. Psychol. 14, 129—135. Bolton, R., 1973. Aggression and hypoglycemia among the Quolla: a study in psychobiological anthropology. Ethology 12, 227—257. Borowsky, I.W., Ireland, M., Resnick, M.D., 2001. Adolescent suicide attempts: risks and protectors. Pediatrics 107, 485—493. Buss, A.H., Perry, M., 1992. The aggression questionnaire. J. Pers. Soc. Psychol. 63, 452—459. Casper, R.C., 1996. Carbohydrate metabolism and its regulatory hormones in anorexia nervosa. Psychiatry Res. 62, 85—96. Centers for Disease Control Prevention, 2004. Suicide attempts and physical fighting among high school students–—United States, 2001. MMWR Morb. Mortal. Wkly. Rep. 53, 474—476. Cryer, P.E., 1993. Glucose counterregulation: prevention and correction of hypoglycemia in humans. Am. J. Physiol. 264, E149—155. Derijks, H.J., Meyboom, R.H., Heerdink, E.R., De Koning, F.H., Janknegt, R., Lindquist, M., Egberts, A.C., 2008. The association between antidepressant use and disturbances in glucose homeostasis: evidence from spontaneous reports. Eur. J. Clin. Pharmacol. 64, 531—538. Escalante Pulido, J.M., Alpizar Salazar, M., 1999. Changes in insulin sensitivity, secretion and glucose effectiveness during menstrual cycle. Arch. Med. Res. 30, 19—22. First, M.B., Gibbon, M., Spitzer, R.L., Williams, J.B.W., Benjamin, L.S., 1998. SCID-II Structured Clinical Interview for DSM IV-Axis II Disorders. Pilgrim Press, Danderyd. Hawton, K., Rodham, K., Evans, E., Weatherall, R., 2002. Deliberate self harm in adolescents: self report survey in schools in England. BMJ 325, 1207—1211. Kumai, M., Tamai, H., Fujii, S., Nakagawa, T., Aoki, T.T., 1988. Glucagon secretion in anorexia nervosa. Am. J. Clin. Nutr. 47, 239—242.
883 McAulay, V., Deary, I.J., Frier, B.M., 2001. Symptoms of hypoglycaemia in people with diabetes. Diab. Med. 18, 690—705. Mitchell, J.E., Bantle, J.P., 1983. Metabolic and endocrine investigations in women of normal weight with the bulimia syndrome. Biol. Psychiatry 18, 355—365. Newcomer, J.W., 2004. Metabolic risk during antipsychotic treatment. Clin. Ther. 26, 1936—1946. Pandit, M.K., Burke, J., Gustafson, A.B., Minocha, A., Peiris, A.N., 1993. Drug-induced disorders of glucose tolerance. Ann. Intern. Med. 118, 529—539. Philipsen, A., Limberger, M.F., Lieb, K., Feige, B., Kleindienst, N., Ebner-Priemer, U., Barth, J., Schmahl, C., Bohus, M., 2008. Attention-deficit hyperactivity disorder as a potentially aggravating factor in borderline personality disorder. Br. J. Psychiatry 192, 118—123. Prochazka, H., Agren, H., 2001. Aggression in the general Swedish population, measured with a new self-rating inventory: the Aggression Questionnaire-Revised Swedish Version (AQ-RSV). Nord. J. Psychiatry 55, 17—23. Russell, J., Hooper, M., Hunt, G., 1996. Insulin response in bulimia nervosa as a marker of nutritional depletion. Int. J. Eat. Disord. 20, 307—313. Tomasik, P.J., Sztefko, K., Starzyk, J., Rogatko, I., Szafran, Z., 2005. Entero-insular axis in children with anorexia nervosa. Psychoneuroendocrinology 4, 364—372. Virkkunen, M., 1982. Reactive hypoglycemic tendency among habitually violent offenders. A further study by means of the glucose tolerance test. Neuropsychobiology 8, 35—40. Virkkunen, M., 1983. Insulin secretion during the glucose tolerance test in antisocial personality. Br. J. Psychiatry 142, 598—604. Virkkunen, M., 1986. Reactive hypoglycemic tendency among habitually violent offenders. Nutr. Rev. 44 (Suppl.), 94—103. Virkkunen, M., Goldman, D., Nielsen, D.A., Linnoila, M., 1995. Low brain serotonin turnover rate (low CSF 5-HIAA) and impulsive violence. J. Psychiatry Neurosci. 20, 271—275. Virkkunen, M., Huttunen, M.O., 1982. Evidence for abnormal glucose tolerance test among violent offenders. Neuropsychobiology 8, 30—34. Virkkunen, M., Narvanen, S., 1987. Plasma insulin, tryptophan and serotonin levels during the glucose tolerance test among habitually violent and impulsive offenders. Neuropsychobiology 17, 19—23. Virkkunen, M., Rawlings, R., Tokola, R., Poland, R.E., Guidotti, A., Nemeroff, C., Bissette, G., Kalogeras, K., Karonen, S.L., Linnoila, M., 1994. CSF biochemistries, glucose metabolism, and diurnal activity rhythms in alcoholic, violent offenders, fire setters, and healthy volunteers. Arch. Gen. Psychiatry 51, 20—27. Virkkunen, M., Rissanen, A., Naukkarinen, H., Franssila-Kallunki, A., Linnoila, M., Tiihonen, J., 2007. Energy substrate metabolism among habitually violent alcoholic offenders having antisocial personality disorder. Psychiatry Res. 150, 287—295. Westrin, A., Frii, K., Traskman-Bendz, L., 2003. The dexamethasone suppression test and DSM-III-R diagnoses in suicide attempters. Eur. Psychiatry 18, 350—355. Winokur, A., Maislin, G., Phillips, J.L., Amsterdam, J.D., 1988. Insulin resistance after oral glucose tolerance testing in patients with major depression. Am. J. Psychiatry 145, 325—330. Wright, J.H., Jacisin, J.J., Radin, N.S., Bell, R.A., 1978. Glucose metabolism in unipolar depression. Br. J. Psychiatry 132, 386— 393. Yasuhara, D., Naruo, T., Nagai, N., Muranaga, T., Nakahara, T., Tanaka, M., Kojima, S., Sagiyama, K., Masuda, A., Inui, A., 2005. Glucose tolerance predicts short-term refeeding outcome in females with anorexia nervosa. Psychosom. Med. 67, 669—676.
a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m
j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / p s y n e u e n
Altered glucose tolerance in women with deliberate self-harm ´n b, Charlotta Sunnqvist a, Lil Tra Sofie Westling a,*, Bo Ahre ¨skman-Bendz a a b
Department of Clinical Sciences, Psychiatry, Lund University, 221 85 Lund, Sweden Department of Clinical Sciences, Medicine, BMC, Lund University, 221 84 Lund, Sweden
Received 23 September 2008; received in revised form 1 December 2008; accepted 30 December 2008
KEYWORDS Glucose tolerance test; Deliberate self-harm; Violent; Insulin; Glucagon; Borderline personality disorder
Summary Disturbances in glucose metabolism are of importance for violent behaviour in men, but studies in women are lacking. We used the 5 h-oral glucose tolerance test (OGTT) in this study of 17 female psychiatric patients, selected for violent behaviour directed against themselves (deliberate self-harm) and 17 healthy controls matched for age and BMI. Following OGTT, patients had higher glucose levels at 30 min ( p = 0.007) and increased glucagon area under the curve ( p = 0.011). Since a co-morbid eating disorder might affect results, we as a post-hoc analysis subgrouped the patients and found that the increased glucagon levels only were present in patients with an eating disorder. In contrast, those without an eating disorder showed a significantly lower p-glucose nadir ( p = 0.015) and unaltered glucagon levels compared to controls. There were no significant differences in insulin and C-peptide levels between patients and controls. We conclude that deliberate self-harm in women may be associated with alterations in carbohydrate metabolism in certain groups. Eating disorder is a confounding factor. # 2009 Elsevier Ltd. All rights reserved.
1. Introduction Several studies have investigated the relationship between glucose metabolism and impulsive aggressive behaviour in healthy men as well as in male forensic psychiatry patients. During the oral glucose tolerance test (OGTT), a reactive hypoglycaemic tendency after the initial rise in plasma (p)glucose has been found in habitually violent men (Bolton, 1973; Benton et al., 1982; Virkkunen and Huttunen, 1982; Virkkunen, 1982, 1983, 1986; Virkkunen and Narvanen, 1987; Virkkunen et al., 1994). In some of these men hypo-
* Corresponding author. Tel.: +46 735 62 60 99; fax: +46 46 17 38 44. E-mail address: [email protected] (S. Westling).
glycaemia has persisted throughout the OGTT. The counterregulation of hypoglycaemia is a complex system, involving a decrease in insulin secretion and an increase in several counter-regulatory factors, augmenting p-glucose, of which the hormone glucagon is the most potent during the immediate phase (Cryer, 1993). The hypoglycaemic tendency found in habitually violent men may thus be explained by either an imbalanced insulin secretion, which does not decrease when p-glucose is reduced, or an insufficient secretion of the counter-regulatory factors raising pglucose. Early studies by Virkkunen and coworkers have found increased insulin levels, proposing this as the cause of the hypoglycaemia (Virkkunen, 1982, 1983). A recent study, examining men with habitually violent behaviour, found that they, compared with healthy controls, had
0306-4530/$ — see front matter # 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.psyneuen.2008.12.015
Altered glucose tolerance in women with deliberate self-harm significantly lower basal glucagon levels, which could be a possible explanation for their hypoglycaemic tendencies (Virkkunen et al., 2007). The men in these studies often showed self-aggression and had made several violent suicide attempts (Virkkunen and Narvanen, 1987). In psychiatric care, violent behaviour among patients is often directed towards the self, and includes deliberate self-harm as well as attempted suicide. This kind of violent behaviour is commonly, but not exclusively, associated with borderline personality disorder, a comorbid diagnose in many of the patients studied by Virkkunen (1986). Furthermore, epidemiological studies among adolescents in the United States have found that aggressive behaviours directed against oneself, and aggressive behaviours towards other people often occur in the same individuals (Borowsky et al., 2001; Centers for Disease Control and Prevention, 2004). Treatment of deliberate self-harm is traditionally psychotherapeutic although hospitalization as well as polypharmacy combining anti-depressant, anti-psychotic as well as tranquilizing medication is common. In patients with severe deliberate self-harm, treatment effect is often limited, and mortality rates in completed suicide are high. Examining glucose metabolism in these patients is a completely new approach and if altered glucose metabolism is found, new treatments might develop as complement to the present ones. Whereas disturbed glucose metabolism has been found in aggressive men, it is not known whether this is the case also in women. We therefore investigated insulin and glucagon secretion and glucose tolerance during OGTT in women with deliberate self-harm. Our primary hypotheses were: (1) Women with deliberate self-harm have high self-rated physical aggression. (2) Women with deliberate self-harm have lower p-glucose nadir (lowest value measured during the OGTT) compared to healthy controls. (3) Women with deliberate self-harm have increased insulin secretion during OGTT compared to healthy controls. (4) Women with deliberate self-harm have decreased glucagon secretion compared to healthy controls.
2. Methods 2.1. Participants From February 2005 to December 2007, female patients with current deliberate self-harm, attending the Division of Psychiatry at the University Hospital in Lund were asked to participate in the study. Inclusion criteria were a history of deliberate self-harm (Hawton et al., 2002) for at least 2 years with at least five incidents occurring during the last 6 months. In order to try to make the group more homogeneous we decided only to include patients with a borderline personality disorder, a condition associated with violent behaviour directed towards self or others. Exclusion criteria were diabetes mellitus or an active liver disease (defined as a twofold rise in p-alanine aminotransferase (ALAT) or p-aspartate aminotransferase (ASAT). Seventeen patients were finally included. Controls matched for sex, age (8 years) and body mass index
879 (BMI) (2 kg m 2) were recruited from a random selection of the population register in the city of Lund and non-randomly among students at the University of Lund. The patients were diagnosed according to the Diagnostic and Statistical Manual of Mental Disorders, fourth edition, axis I, II and III (American Psychiatric Association, 2000). All subjects were classified and rated by the corresponding author. All patients had a borderline personality disorder according to the Structured Clinical Interview for DSM IVAxis II disorders (First et al., 1998). Two patients fulfilled the criteria for one axis I diagnosis, six fulfilled the criteria for two, further six fulfilled the criteria for three, and three patients fulfilled the criteria for four axis I diagnoses. Twelve of the patients had anxiety disorders, 10 had a mood disorder, seven had an eating disorder, seven had a substance use disorder (four with alcohol abuse), two had an Attention Deficit Hyperactivity Disorder, one had a dissociative disorder, and one had psychotic symptoms not other specified. Thirteen patients had somatic diagnoses, of which asthma/ allergy (n = 6) and hypothyroidism (n = 2) were the most common ones. All patients used medication. Fifteen used a psychotropic medication; no patient used lithium. The most common medication for somatic diseases were inhalators/ anti-allergic medication (n = 6), contraceptives (n = 5) and omeprazol (n = 4). Six of the healthy controls used hormonal contraceptives and five used nutritional supplements. The patients’ medications had not been changed for a median of 33 days (interquartile range 24—88 days). Only nine of the 17 patients had regular menstruation.
2.2. The Aggression-Questionnaire Revised Swedish Version (AQ-RSV) Level of aggression was estimated by use of the AQ-RSV (Prochazka and Agren, 2001). AQ-RSV consists of 29 items, distributed into four scales, measuring the aggression factors physical aggression (nine items), verbal aggression (five items), anger (seven items) and hostility (eight items). Each item is rated on a 4-point scale, from the least to the most characteristic of the subject. We used the scale for Physical aggression in order to see if our subjects could be regarded as comparable to the habitually violent men in the studies by Virkkunen and co-workers. The raw score presented in the table is the total sum of the ratings of items included in the scale for physical aggression. In the English version of the aggression questionnaire, the sub-scale physical aggression has a reliability of 0.80 and significant correlations have been found between the sub-scale and peer nomination (Buss and Perry, 1992). Reliability and validity has not been tested in the Swedish version. However, the internal consistency of the four aggression subscales and the total score was evaluated by the alpha coefficients (Cronbach’s alpha). The alphas of physical aggression were comparable in the American and Swedish populations and the alpha coefficients indicated considerable internal consistency. Normative data for AQRSV has been obtained earlier from the general population (Prochazka and Agren, 2001). We used a formula developed by Helena Prochazka (personal communication) to compare our patients to women in the general Swedish population. AQ-RSV has not previously been used on patients with deliberate self-harm.
880
S. Westling et al.
2.3. Oral glucose tolerance test and sampling The subjects arrived after one night of fasting. Blood samples were examined for possible infections, anaemia, hepatic and thyroid diseases, plasma-lipids and electrolytes. Test samples included glucose, insulin, c-peptide and glucagon. After baseline sampling the subjects ingested 300 ml solution-containing water mixed with 75 g glucose. New plasma samples were drawn after 15, 30, 45, 60, 75, 90, 120, 150, 180, 240 and 300 min. Subjects were asked to report any adverse symptoms and additional samples for immediate analysis of capillary blood glucose were drawn. The test samples were collected in tubes containing EDTA (7.4 mmol/l; final concentration) and aprotinin (500 kallikrein inhibitor units/ ml blood; Novo Nordisk, Bagsvaerd, Denmark). 10,000 IE trasylol/ml blood was added to the tube collected for glucagon analysis. All test samples were immediately put on ice and centrifuged at +4 8C, 3000 rpm, within at most 3 h. Plasma was frozen at 80 8C until analysis. Insulin, C-peptide and glucagon concentrations were analyzed with double antibody radioimmunoassay techniques (Linco Research, St. Charles, MO); glucose was determined with the glucose oxidase method.
2.4. Statistical analyses A power analysis was made before the study was initiated. We used p-glucose nadir as a primary effect variable, which in former studies have shown a standard deviation of 0.5. With a clinically relevant difference of 0.5 mM and power of 0.8, a significance level of 5% demands a sample of 17 matched couples. The software SPSS 13.0 for Windows was Table 1
used for all statistical calculations. Because of the small sample size we used non-parametric statistics. p-Values were calculated with Mann—Whitney U-test. Results are presented as median and interquartile range. BMI was calculated by dividing the patients’ weight in kg by the squared height. Area under curve (AUC) was calculated with the trapezoid method. As a measure of insulin sensitivity, we used insulinogenic index defined as (insulin30 min insulin0 min) (glucose30 min glucose0 min) 1 (Ahren et al., 2008). One patient had a missing value for glucagon and one had a missing value for c-peptide during the glucose tolerance test. When calculating the AUC for glucagon and c-peptide, respectively in these patients, the missing values were estimated to be the mean of the values immediately preceding and following the missing one. One patient had missing values for the last observation (300 min). This patient only completed 240 min of the protocol due to severe suicidal impulses, increasing during the test. The values for the sample at 300 min were then calculated by dividing the remaining 33 subjects’ values for 300 min by the values for 240 min and then multiplying the quotient with the patient’s observed values at 240 min. The quotients were all normally distributed, within a reasonable range. Two patients had each one missing value on an item in the AQ-RSV. The score of this item was then calculated as the median of the scores from the remaining eight items.
2.5. Ethical approval The study was approved by the Regional Ethical Review Board in Lund and all subjects gave written informed consent before the study.
Comparisons between patients and healthy controls (median and interquartile range).
Age (years) BMI (kg m 2) p-GT (glutamyltransferase) (ref < 0.4 mkat/l) p-ALAT (alanine aminotransferase) (ref < 0.7 mkat/l) fp-Glucose (mM) p-Glucose 30 min (mM) p-glucose 300 min (mM) p-Glucose nadir (mM) p-Glucose AUC (M min) fp-Insulin (pM) p-Insulin 30 min (pM) p-Insulin AUC (mM min) Insulinogenic index 30 min (pmol mmol 1) a fp-c-Peptide (nM) p-c-Peptide 30 min (nM) p-c-Peptide AUC (nM min) fp-Glucagon (ng l 1) p-Glucagon 30 min (ng l 1) p-Glucagon AUC (mg l 1) AQ-RSV, physical aggression score
Patients (n = 17)
Controls (n = 17)
p
20 (19—28) 22.1 (18.8—25.8) 0.26 (0.20—0.32) 0.25 (0.22—0.37) 4.7 (4.1—5.2) 8.2 (7.2—9.2) 4.2 (3.7—4.6) 3.2 (2.8—3.6) 1.6 (1.4—1.7) 70 (43—90) 526 (361—738) 67 (39—100) 106 (81—180) 0.41 (0.34—0.86) 2.14 (1.44—3.26) 451 (276—586) 69 (46—86) 60 (50—78) 16 (15—23) 23 (16—28)
24 (21—27) 21.1 (19.1—25.2) 0.19 (0.16—0.26) 0.25 (0.20—0.38) 4.6 (4.2—5.1) 7.0 (5.6—7.6) 4.6 (4.0—4.8) 3.9 (2.7—4.2) 1.5 (1.3—1.7) 80 (61—85) 471 (301—826) 67 (45—86) 207 (99—327) 0.54 (0.40—0.75) 2.35 (1.52—3.5) 456 (321—528) 46 (37—66) 46 (37—65) 13 (10—15) 11 (9—12)
0.218 0.892 0.110 0.817 0.946 0.007 ** 0.204 0.231 0.160 0.432 0.812 0.973 0.377 0.766 0.786 0.865 0.073 0.053 0.011 * <0.001
Abbreviations: BMI: body mass index; fp: fasting plasma; p: plasma; AQ-RSV: Aggression Questionnaire Revised Swedish Version. a Insulinogenic index = (insulin30 min insulin0 min) (glucose30 min glucose0 min) 1 (Ahren et al., 2008). * p = 0.011, Mann—Whitney U-test. ** p = 0.007, Mann—Whitney U-test.
Altered glucose tolerance in women with deliberate self-harm
3. Results 3.1. Patients (n = 17) and controls (n = 17) (Table 1) 3.1.1. Basal characteristics There were no significant differences between patients and controls in BMI, age, p-glutamyltransferase (GT) or p-ALAT. 3.1.2. AQ-RSV Patients had significantly higher scores on physical aggression (23 [16—28]) compared to controls (11 [9—12]) ( p < 0.001). When relating to women in the general Swedish population, our patients had higher scores with a ratio of 1.45:1 (1.04— 1.74:1). 3.1.3. Glucose There was no significant difference between patients and controls in fasting plasma (fp-) glucose. Following the glucose administration, glucose levels increased to a peak after 30 (30—45) min; and thereafter glucose levels fell. After 180 (120—180) min, a nadir was reached. Patients had significantly higher p-glucose after 30 min (8.2 [7.2—9.2] mM) compared to controls (7.0 [5.6—7.6] mM) ( p = 0.007). There were no significant differences in p-glucose nadir, p-glucose at 300 min or in glucose AUC between patients and controls. There was no significant correlation between p-glucose nadir and scores on physical aggression. 3.1.4. Insulin There were no significant differences in fp-insulin, p-insulin 30 min or insulin AUC between patients and controls. No significant difference was found between patients and controls in insulin sensitivity measured as insulinogenic index. 3.1.5. C-peptide There were no significant differences between patients and controls in fp-c-peptide, p-c-peptide 30 min or c-peptide AUC. 3.1.6. Glucagon There were no significant differences between patients and controls in fp-glucagon or p-glucagon 30 min although a trend could be seen towards increased 30 min levels in patients (60 [50—78] ng l 1) compared to controls (46 [37—65] ng l 1) ( p = 0.053). Patients had a significantly increased glucagon AUC (16 [15—23] mg min/l) compared to controls (13 [10— 15] mg min/l) ( p = 0.011).
3.2. Patients without an eating disorder (n = 10) and controls (n = 10) Patients without an eating disorder showed significantly increased scores on AQ-RSV Physical aggression (24 [20— 29]) compared to controls (11 [9—12]) ( p = 0.001). These patients had significantly lower p-glucose nadir (3.2 [2.8— 3.6] mM) than controls (4.0 [3.4—4.2] mM) ( p = 0.015). They also had lower p-glucose at the endpoint (300 min) (4.0 [3.3— 4.6] mM) than controls (4.7 [4.3—5.2] mM) ( p = 0.029). There was no significant difference between patients and controls in p-glucose 30 min or glucagon AUC. None of the other
881 parameters measured showed any significant differences between patients and controls. There was no significant correlation between p-glucose nadir and scores on physical aggression.
3.3. Patients with an eating disorder (n = 7) and controls (n = 7) Patients with an eating disorder showed significantly increased scores on AQ-RSV physical aggression (22 [14— 24]) compared to controls (11 [9—12]) ( p = 0.012). These patients showed significantly increased glucagon AUC (17 [16—29] mg min/l) compared to controls (13 [12— 21] mg min/l) ( p = 0.038). They also showed a trend towards increased p-glucose 30 min (7.8 [7.2—9.4] mM) compared to controls (7.0 [5.0—7.4] mM) ( p = 0.053). None of the other parameters measured showed any significant differences between patients and controls.
4. Discussion We have investigated glucose, insulin, c-peptide and glucagon in plasma during OGTT and found increased glucagon levels and increased p-glucose after 30 min in women with deliberate self-harm compared to healthy controls. We also investigated all subjects with a self-rating scale for physical aggression and found significantly increased scores in patients. Although no study has investigated a correlation between observed violent behaviour and high ratings on physical aggression on the AQ-RSV, the high scores obtained by our patients is an indicator that this sample might be somewhat comparable to the habitually violent men studied by Virkkunen and co-workers (Virkkunen and Huttunen, 1982; Virkkunen, 1982, 1983, 1986; Virkkunen and Narvanen, 1987; Virkkunen et al., 1994). After the initial calculations, and due to heterogeneity of the patient population, we performed post hoc analyses to identify factors, which could explain differences in subgroups. We thus analysed individual curves for the patients and controls, as well as at groups of curves including patients with specific diagnoses, which we thought might confound our results (i.e. eating disorders, MDD, substance abuse). We found that patients with an eating disorder had elevated glucagon curves. This was not observed in other groups. As a post hoc analysis we therefore separately analysed the results from patients with versus without an eating disorder. Our finding of an increased p-glucagon AUC was confirmed only in patients with an eating disorder. These patients further showed a trend towards increased p-glucose at 30 min. We consider this trend to be subsequent to the rise in glucagon secretion, since the whole glucose AUC seemed elevated in these patients compared to controls, however not significantly. Among patients without an eating disorder, we found a low p-glucose nadir when compared with controls. Our primary hypothesis was therefore confirmed, but only in this subgroup of patients. In addition, patients not suffering from an eating disorder had lower p-glucose than controls at endpoint during OGTT. Thus, regardless of co-morbidity and concomitant medication, women with deliberate self-harm without an eating disorder showed changes in glucose metabolism, com-
882 parable to those found in men with habitually violent behaviour (Virkkunen and Huttunen, 1982; Virkkunen, 1982, 1983, 1986; Virkkunen and Narvanen, 1987; Virkkunen et al., 1994). Aggression is one out of several symptoms of hypoglycaemia (McAulay et al., 2001). Our patients without an eating disorder had a p-glucose nadir of 3.2 mM, which is significantly lower compared to 4.0 mM in controls. This difference is of approximately the same magnitude as earlier found in men, but absolute glucose levels are higher in our patients as well as in our controls (Virkkunen and Huttunen, 1982; Virkkunen, 1982, 1986; Virkkunen et al., 1994). The cause of this difference between absolute values in our study and the Virkkunen ones is not known but is intriguing since not only the patients, but also controls showed higher levels in our study. We did not find any significant correlation between glucose nadir and the ratings for Physical aggression. A causal relation between the lower p-glucose and the aggressive behaviour seem less likely and a common mechanism behind both cannot be excluded. Epinephrine, growth hormone as well as cortisol are factors involved in glucose counterregulation (Cryer, 1993) but not measured in this study. Low pre- and postdexamethasone cortisol has been found in suicide attempters with cluster B personality disorder compared to suicide attempters and healthy controls (Westrin et al., 2003). Furthermore cortisol as well as epinephrine participates in mood regulation. Alterations in these substances might explain the difference found in p-glucose nadir between our patients and controls. However, glucagon is known to be the most potent counter-regulating factor, which means that a normal secretion of glucagon should have prevented hypoglycaemia (Cryer, 1993). Virkkunen et al. (1995) have proposed serotonin as a common denominator, as it might affect behaviour as well as glucose homeostasis. Low concentrations of the serotonin metabolite 5-hydroxyindole acetic acid in cerebrospinal fluid has furthermore been found in impulsive violent men prone to hypoglycaemia (Virkkunen et al., 1994). These are thus substances of interest for future research. Our hypothesis of enhanced insulin secretion in patients could not be verified. Nor did we find any significant differences in insulin sensitivity, explaining the low p-glucose nadir. In studies investigating male forensic patients, results have been somewhat divergent. Enhanced p-insulin during OGTT was found in offenders with intermittent explosive disorder, but not in men with violent antisocial personality disorder (Virkkunen, 1986). The studies on male offenders differ from ours, since the offenders were imprisoned during 5 months before examination and thus abstinent from alcohol and drugs and with a reasonably controlled intake of food and psychotropic drugs. Another difference is that in our study, but not in the studies by Virkkunen et al., not only p-insulin but also c-peptide was measured. C-peptide, in contrast to insulin, is not subjected to a substantial uptake by the liver and thus more relevant when estimating insulin secretion. However, since our study lacked control over a number of mechanisms affecting glucose metabolism, a true difference in insulin sensitivity cannot be excluded. Another important finding in this study is that patients with an eating disorder had an increased glucagon secretion and a trend towards increased p-glucose 30 min, in contrast to patients without an eating disorder. The elevated glucagon
S. Westling et al. secretion would serve to raise p-glucose, thus explaining the differences in p-glucose 30 min as well as in p-glucose nadir between the two subgroups. Eating disorder per se is however a less likely explanation for the glucagon secretion (Mitchell and Bantle, 1983; Kumai et al., 1988; Casper, 1996; Russell et al., 1996; Tomasik et al., 2005; Yasuhara et al., 2005). The most resonable explanation for the elevated glucagon secretion is the lack of control over dietary intake in our study. The irregular and inadequate meals compatible with an eating disorder probably affects and confounds these results. Our patient group was homogeneous for deliberate selfharm and borderline personality disorder, but heterogeneous for a number of factors including concomitant diagnoses, medication, intake of illegal drugs and alcohol, diet and menstruation. This is an inevitable consequence of the study subject; deliberate self-harm, which is a cross-diagnostic behaviour of major clinical relevance. Concerning co-morbidity, our participants well represent patients with deliberate self-harm and borderline personality disorder in a naturalistic setting (Philipsen et al., 2008). Our study thus approximates real life better than it would have done with more stringent exclusion criteria. Another limitation is that all patients, but one, were outpatients. This makes the circumstances during the days preceding the test less controllable. Concurrent medication was prevalent but relatively stable. Some anti-psychotic as well as anti-depressant medication can induce obesity, with subsequent insulin resistance and type 2 diabetes mellitus (T2DM) (Newcomer, 2004; Derijks et al., 2008). Oral contraceptives are also related to an increased risk of impaired glucose tolerance (Pandit et al., 1993). However, we matched controls for BMI and none of the subjects showed either insulin resistance or T2DM. The co-occurrence of diseases other than eating disorder that might affect glucose metabolism is a further limitation. Major depressive disorder (MDD) is associated with insulin resistance (Wright et al., 1978; Winokur et al., 1988). Six of our patients suffered from MDD but none of them showed insulin resistance. Four of the patients were diagnosed with alcohol abuse. However, no participants showed increased levels of p-carbohydrate deficient transferrin (CDT) and active liver disease (as defined in methods) was an exclusion criterion. We have thus tried to minimize the effect of a comorbid alcohol abuse. Another limitation is the influence of the menstrual cycle on glucose tolerance (Escalante Pulido and Alpizar Salazar, 1999). Since only 9 of the 17 patients had regular menstruation we could not to co-ordinate the patients’ menstrual cycles with those of the controls. In conclusion our results seemingly confirm and extend findings by others of alterations in carbohydrate metabolism in persons with violent behaviour, here mainly shown as deliberate self-harm. We suggest that impulsive aggression in female patients is associated with a deviant carbohydrate metabolism with impaired glucagon response to reactive glucose nadir after oral glucose challenge and that a combined eating disorder restores this by enhancing glucagon secretion.
Role of funding source The work was supported by grants from the Swedish Research Council no. 14548 and 6834, Lund University Hospital Research Funds, the Faculty of Medicine, Lund University,
Altered glucose tolerance in women with deliberate self-harm the So ¨derstro ¨m-Ko ¨nigska Foundation, the Sjo ¨bring Foundation, the OM Persson Foundation, the Lundbeck Foundation, Swedish Psychiatry Foundation and Region Ska ˚ne. None of the funding organisations had any further role in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication.
Conflict of interest All authors declare that they have no conflicts of interest.
Acknowledgements The authors are grateful to Lilian Bengtsson and Kristina Andersson for expert technical assistance in the analyses of samples, to Helena Prochazka for supplying profound knowledge on the AQ-RSV and to Hans Bendz for important comments on the manuscript.
References Ahren, B., Pratley, R.E., Soubt, M., Dunning, B.E., Foley, J.E., 2008. Clinical measures of islet function: usefulness to characterize defects in diabetes. Curr. Diab. Rev. 4, 129—145. American Psychiatric Association, 2000. Diagnostic and Statistical Manual of Mental Disorders fourth edition. American Psychiatric Press, Washington, DC. Benton, D., Kumari, N., Brain, P.F., 1982. Mild hypoglycaemia and questionnaire measures of aggression. Biol. Psychol. 14, 129—135. Bolton, R., 1973. Aggression and hypoglycemia among the Quolla: a study in psychobiological anthropology. Ethology 12, 227—257. Borowsky, I.W., Ireland, M., Resnick, M.D., 2001. Adolescent suicide attempts: risks and protectors. Pediatrics 107, 485—493. Buss, A.H., Perry, M., 1992. The aggression questionnaire. J. Pers. Soc. Psychol. 63, 452—459. Casper, R.C., 1996. Carbohydrate metabolism and its regulatory hormones in anorexia nervosa. Psychiatry Res. 62, 85—96. Centers for Disease Control Prevention, 2004. Suicide attempts and physical fighting among high school students–—United States, 2001. MMWR Morb. Mortal. Wkly. Rep. 53, 474—476. Cryer, P.E., 1993. Glucose counterregulation: prevention and correction of hypoglycemia in humans. Am. J. Physiol. 264, E149—155. Derijks, H.J., Meyboom, R.H., Heerdink, E.R., De Koning, F.H., Janknegt, R., Lindquist, M., Egberts, A.C., 2008. The association between antidepressant use and disturbances in glucose homeostasis: evidence from spontaneous reports. Eur. J. Clin. Pharmacol. 64, 531—538. Escalante Pulido, J.M., Alpizar Salazar, M., 1999. Changes in insulin sensitivity, secretion and glucose effectiveness during menstrual cycle. Arch. Med. Res. 30, 19—22. First, M.B., Gibbon, M., Spitzer, R.L., Williams, J.B.W., Benjamin, L.S., 1998. SCID-II Structured Clinical Interview for DSM IV-Axis II Disorders. Pilgrim Press, Danderyd. Hawton, K., Rodham, K., Evans, E., Weatherall, R., 2002. Deliberate self harm in adolescents: self report survey in schools in England. BMJ 325, 1207—1211. Kumai, M., Tamai, H., Fujii, S., Nakagawa, T., Aoki, T.T., 1988. Glucagon secretion in anorexia nervosa. Am. J. Clin. Nutr. 47, 239—242.
883 McAulay, V., Deary, I.J., Frier, B.M., 2001. Symptoms of hypoglycaemia in people with diabetes. Diab. Med. 18, 690—705. Mitchell, J.E., Bantle, J.P., 1983. Metabolic and endocrine investigations in women of normal weight with the bulimia syndrome. Biol. Psychiatry 18, 355—365. Newcomer, J.W., 2004. Metabolic risk during antipsychotic treatment. Clin. Ther. 26, 1936—1946. Pandit, M.K., Burke, J., Gustafson, A.B., Minocha, A., Peiris, A.N., 1993. Drug-induced disorders of glucose tolerance. Ann. Intern. Med. 118, 529—539. Philipsen, A., Limberger, M.F., Lieb, K., Feige, B., Kleindienst, N., Ebner-Priemer, U., Barth, J., Schmahl, C., Bohus, M., 2008. Attention-deficit hyperactivity disorder as a potentially aggravating factor in borderline personality disorder. Br. J. Psychiatry 192, 118—123. Prochazka, H., Agren, H., 2001. Aggression in the general Swedish population, measured with a new self-rating inventory: the Aggression Questionnaire-Revised Swedish Version (AQ-RSV). Nord. J. Psychiatry 55, 17—23. Russell, J., Hooper, M., Hunt, G., 1996. Insulin response in bulimia nervosa as a marker of nutritional depletion. Int. J. Eat. Disord. 20, 307—313. Tomasik, P.J., Sztefko, K., Starzyk, J., Rogatko, I., Szafran, Z., 2005. Entero-insular axis in children with anorexia nervosa. Psychoneuroendocrinology 4, 364—372. Virkkunen, M., 1982. Reactive hypoglycemic tendency among habitually violent offenders. A further study by means of the glucose tolerance test. Neuropsychobiology 8, 35—40. Virkkunen, M., 1983. Insulin secretion during the glucose tolerance test in antisocial personality. Br. J. Psychiatry 142, 598—604. Virkkunen, M., 1986. Reactive hypoglycemic tendency among habitually violent offenders. Nutr. Rev. 44 (Suppl.), 94—103. Virkkunen, M., Goldman, D., Nielsen, D.A., Linnoila, M., 1995. Low brain serotonin turnover rate (low CSF 5-HIAA) and impulsive violence. J. Psychiatry Neurosci. 20, 271—275. Virkkunen, M., Huttunen, M.O., 1982. Evidence for abnormal glucose tolerance test among violent offenders. Neuropsychobiology 8, 30—34. Virkkunen, M., Narvanen, S., 1987. Plasma insulin, tryptophan and serotonin levels during the glucose tolerance test among habitually violent and impulsive offenders. Neuropsychobiology 17, 19—23. Virkkunen, M., Rawlings, R., Tokola, R., Poland, R.E., Guidotti, A., Nemeroff, C., Bissette, G., Kalogeras, K., Karonen, S.L., Linnoila, M., 1994. CSF biochemistries, glucose metabolism, and diurnal activity rhythms in alcoholic, violent offenders, fire setters, and healthy volunteers. Arch. Gen. Psychiatry 51, 20—27. Virkkunen, M., Rissanen, A., Naukkarinen, H., Franssila-Kallunki, A., Linnoila, M., Tiihonen, J., 2007. Energy substrate metabolism among habitually violent alcoholic offenders having antisocial personality disorder. Psychiatry Res. 150, 287—295. Westrin, A., Frii, K., Traskman-Bendz, L., 2003. The dexamethasone suppression test and DSM-III-R diagnoses in suicide attempters. Eur. Psychiatry 18, 350—355. Winokur, A., Maislin, G., Phillips, J.L., Amsterdam, J.D., 1988. Insulin resistance after oral glucose tolerance testing in patients with major depression. Am. J. Psychiatry 145, 325—330. Wright, J.H., Jacisin, J.J., Radin, N.S., Bell, R.A., 1978. Glucose metabolism in unipolar depression. Br. J. Psychiatry 132, 386— 393. Yasuhara, D., Naruo, T., Nagai, N., Muranaga, T., Nakahara, T., Tanaka, M., Kojima, S., Sagiyama, K., Masuda, A., Inui, A., 2005. Glucose tolerance predicts short-term refeeding outcome in females with anorexia nervosa. Psychosom. Med. 67, 669—676.