| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Psychiatry (P.M., A.F., E.C., M.M.), University of Naples SUN, Naples, Italy; the Department of Neurology and Psychiatry (P.S., A.F.), University of Padua, Padua, Italy; the Department of Psychiatry, Neurobiology, Pharmacology and Biotechnologies (M.M., A.R.), University of Pisa, Pisa, Italy; and the Istituto Scientifico Ospedale San Raffaele (L.B., S.E.), Vita-Salute University, Faculty of Psychology, Milan, Italy.
Address correspondence and reprint requests to Palmiero Monteleone, MD, Department of Psychiatry, University of Naples S.U.N., Largo Madonna delle Grazie, 80138 Naples, Italy. E-mail: monteri{at}tin.it
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Methods: A total of 219 white women (125 bulimics and 94 healthy control subjects) underwent a blood sample collection for 5HTTLPR genotyping and a clinical evaluation assessing comorbidity for axis I and II psychiatric disorders, harm avoidance personality dimension, and body composition (only patients).
Results: The distribution of the 5HTTLPR genotypes did not significantly differ between patients and control subjects, although the L allele was significantly more frequent in the former. Bulimic individuals carrying at least one copy of the S allele had significantly lower mean body mass index and body fat mass values and significantly higher mean harm avoidance score than patients with the LL genotype. No significant association was found between the 5HTTLPR genotype and comorbid axis I and II psychiatric disorders.
Conclusions: These findings support the view that polymorphic variants of the 5HTT promoter region do not play a part in predisposing to BN, whereas they seem to predispose bulimic individuals to nutritional impairment and increased harm avoidance.
Key Words: bulimia nervosa • serotonin transporter • harm avoidance • polymorphism • genetics • comorbidity
Abbreviations: ANOVA = analysis of variance; BMI = body mass index; BW = body weight; BN = bulimia nervosa; MINI = Mini International Neuropsychiatric Interview; NS = not significant; 5HT = serotonin; 5HTT = serotonin transporter; 5HTTLPR = serotonin transporter length polymorphic region; SCID-IP = Structured Clinical Interview for DSM IV Axis I disorders–Patient Edition; SCID-II = Structured Clinical Interview for DSM IV Axis II personality disorders; TCI-R = Temperament and Character Inventory Revised.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Several studies have uncovered derangements of 5HT transmission in both symptomatic and recovered bulimic patients (10–13). If 5HT aberrations underlie the biologic vulnerability to BN development, then an association of particular genes involved in serotonergic transmission and BN should be expected. Recently, a single photon emission computed tomography study (14) indicated reduced hypothalamic and thalamic 5HT transporter (5HTT) availability in BN. This finding and the antibulimic efficacy of the selective 5HT reuptake inhibitors (15), which act specifically on 5HTT, suggest the 5HTT gene as a good candidate for genetic studies of BN.
The human 5HTT gene is located on chromosome 17q11.1 to 17q.12. A polymorphism in the promoter region of the 5HTT gene (5HTTLPR) has been described (16). It consists of a 44 base-pair deletion (short or S variant) or insertion (long or L variant), which is endowed with functional consequences because the S form is associated with a lower transcriptional activity and a reduced 5HT reuptake efficiency than the L isoform (17). To the best of our knowledge, so far, only three studies (18–20) have assessed the role of 5HTTLPR as a susceptibility factor for BN. Two of these studies (19,20) found no significant association between 5HTTLPR and BN, whereas the remaining one (18) found a significant association between the S allele and BN. Therefore, the possibility that 5HTTLPR plays a role in the genetic vulnerability to BN is still an open question. Moreover, the probability exists that in BN, 5HTTLPR is associated with personality traits related to 5HT dysfunction such as harm avoidance and/or represents a vulnerability factor for psychiatric comorbidity, including mood, anxiety, and personality disorders, which frequently co-occur with BN (21).
To explore these issues, we designed a case–control study aiming to investigate 1) differences in 5HTTLPR genotype frequencies between patients with BN and healthy comparison subjects; 2) differences in the harm avoidance personality dimension between patients and control subjects; 3) the possible association of harm avoidance with the 5HTTLPR genotype; 4) the relationships between 5HTTLPR and nutritional parameters, axis I or II psychiatric comorbidities in patients with BN.
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
A total of 125 female white patients (age range, 18–44 years) were enrolled in the study: 64 were recruited in Naples (South Italy), 24 in Padua (Nord-Est Italy), 21 in Milan (North Italy), and 14 in Pisa (Middle Italy). All of them met the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) diagnosis of BN purging subtype, with self-induced vomiting as the main compensatory behavior, as confirmed by the Structured Clinical Interview for DSM IV-Patient Edition (SCID-IP) (22). They underwent the following clinical assessments: 1) axis I comorbid psychiatric disorders were investigated using the SCID-I-P; 2) comorbid personality disorders were evaluated by means of the SCID-II for DSM-IV (23) (given our population, we disregarded the borderline personality disorder criterion referring to "overeating"); and 3) harm avoidance personality dimension was assessed by the Temperament and Character Inventory Revised (TCI-R) of Cloninger (24). Finally, all patients underwent a structured clinical interview aiming to collect historical and clinical data concerning their bulimic symptomatology.
A group of 94 drug-free white healthy women (age range, 18–41 years) were also recruited: 61 were enrolled in Naples, 13 in Padua, 11 in Milan, and nine in Pisa. They were mentally healthy as assessed by the Mini International Neuropsychiatric Interview (MINI) (25). Information on eating habits was collected by a clinical interview; harm avoidance personality dimension was assessed by the TCI-R.
All of the subjects provided written informed consent to participate in the study after a complete description of the study procedure.
Diagnostic and psychopathologic assessments were made by investigators (psychiatrists and psychologists) who were formally trained in the use of the SCID-I, SCID-II, and MINI. This training included watching videotapes of a master interviewer administering SCID-I, SCID-II, and MINI; on-site training by master interviewers; a series of tryout interviews of patient subjects with all raters making independent ratings and later discussing sources of disagreement. This was followed by all interviewers independently scoring a series of three additional SCID-I, SCID-II, and MINI videotapes. At the end of training, there was no disagreement among investigators (Cohen’s kappa = 1) in classifying one patient with anorexia nervosa, restricting subtype, one patient with BN, one patient with schizophrenia, and three patients with personality disorders.
In each subject, the body mass index (BMI) was calculated as the ratio between body weight (BW) (kg) and height (m2); moreover, in patients, the body composition was evaluated by means of a bioelectrical impedance analyser (STA/BIA, Akern Srl, Florence, Italy).
The study was approved by the local ethic committees of the four centers.
Genomic DNA was extracted from whole blood and target DNA was amplified by polymerase chain reaction; the 5HTTLPRgenotype was determined as described by Deckert et al (26).
Statistics
Statistical analyses were performed using the BMDP statistic software package (27).
Association tests were performed by using the 
2 test for comparison of genotype frequencies between patients and healthy control subjects and of axis I and II psychiatric comorbidities among genotypes in the patient group. Moreover, to test whether the differences in nutritional parameters and harm avoidance between the two groups were influenced by 5HTTLPR genotype, a two-way analysis of variance (ANOVA) was performed with diagnosis (patients with BN versus healthy women) and 5HTTLPR genotype group (SS versus SL versus LL) as independent variables. Post hoc one-way ANOVAs were performed to assess statistical differences among the various genotypes separately for the two groups. These analyses were performed two times: the first one in a three-genotype model (SS versus SL versus LL genotypes) and the second one in a two-genotype model in which SS and SL genotypes were combined together according to a dominant genetic paradigm (28).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
2 = 3.676, df = 6, p = not significant [NS] for patients; 2 = 4.941, df = 6, p = NS). The distribution of the 5HTTLPR genotypes did not differ significantly from that expected according to the Hardy-Weinberg equilibrium in both patients (2 = 3.77; df = 1, p = NS) and control subjects (2 = 2.37; df = 1, p = NS). No significant differences were detected in 5HTTLPR genotype distributions between comparison subjects and patients with BN (2 = 4.78; df = 2, p = NS). A statistically significant distribution, instead, emerged in the allele frequencies among the two groups (2 = 5.62; df = 1, p = .01) with the L allele being more frequent in BN subjects than in control women (Table 1); the odds ratio for the L allele was 1.58 (95% confidence intervals, 1.06–2.31).
|
Statistical comparisons of BMI values by means of two-way ANOVA revealed no main effects of diagnosis (F1, 213 = 1.81, p = NS) and of 5HTTLPR genotype (F2, 213 = 1.03, p = NS) but a significant diagnosis x genotype interaction (F2, 213 = 5.39, p = .0052). Post hoc ANOVAs with genotype as an independent variable showed that among women with BN, genotype had a significant effect on BMI values (F2, 122 = 5.06; p = .0077) with SS and SL subjects exhibiting lower values of mean BMI than carriers of the LL genotype (F1, 72 = 8.64, p = .004 and F1, 93 = 6.11, p = .01, respectively). There was no effect of genotype on BMI in comparison subjects (F2, 91 = 2.06; p = NS). One-way ANOVA on body fat and body lean mass values in patients with BN disclosed a significant effect of genotype on body fat mass (F2, 122 = 4.07; p = .01) but not on body lean mass (F2, 122 = 0.46; p = NS). SS subjects exhibited lower values of mean body fat mass than carriers of the LL genotype (F1, 72 = 7.57; p = .007); patients with the SL genotype, instead, showed only a trend toward a significant decreased mean body fat mass value as compared with LL individuals (F1, 93 = 2.78; p = .09). These results did not change when SS and SL subjects were combined together and analyzed against LL patients; indeed, women with BN carrying at least one copy of the S allele had mean BMI and body fat mass values significantly lower than those carrying the LL genotype (F1, 123 = 8.75, p = .0037, and F1, 123 = 6.38; p = .01, respectively) (Table 2).
|
Regarding harm avoidance scores, two-way ANOVA showed a significant main effect of diagnosis (F1, 213 = 75.84, p < .0000001) and of 5HTTLPR genotype (F2, 213 = 5.13, p = .024) with no significant diagnosis x genotype interaction (F2, 213 = 1.84, p = .1). Anyway, although the diagnosis x genotype interaction did not reach statistical significance, it seemed that the effect of genotype on harm avoidance in patients with BN was larger than in control subjects. Indeed, post hoc ANOVAs with genotype as an independent variable showed an effect of genotype on harm avoidance in women with BN (F2, 122 = 4.11; p = .018) with carriers of the SS genotype or SL genotype exhibiting higher mean values of harm avoidance than carriers of the LL genotype (F1, 72 = 5,87, p = .01 and F1, 93 = 5, 24, p = .02, respectively); there was no effect of genotype on harm avoidance in comparison subjects (F2,91 = 1.80, p = NS). These results did not change when SS and SL subjects were combined together and analyzed against LL patients; indeed, women with BN carrying at least one copy of the S allele had a mean harm avoidance score significantly higher than those carrying the LL genotype (F1, 123 = 7.87, p = .0058) (Table 2).
To test whether higher harm avoidance scores mediated the associations between the S allele and lower BMI/body fat mass in bulimics, correlation analyses between harm avoidance scores and BMI or body fat mass values were performed. These analyses showed that harm avoidance was not significantly correlated to both BMI (r = –0.13, p = .1) and body fat mass values (r = –0.01; p = .8), thus excluding a role of harm avoidance as mediator of genotype effects on BMI and body fat mass in BN.
Seventy-one patients (56.8%) had at least one lifetime comorbid axis I disorder (55 patients had an affective disorder, 37 had an anxiety disorder, and 21 had anorexia nervosa). The distribution of patients with or without axis I comorbid disorders did not significantly differ among the three genotype groups (2 = 0.17, df = 2, p = NS). When the analysis was focused on the presence of a history of anorexia nervosa, the distribution of patients with or without anorexia nervosa did not significantly differ among the three genotype groups (2 = 0.57, df = 2, p = NS). Fifty-four patients (43.2%) had at least one comorbid axis II disorder (eight had a cluster A disorder, 35 had a cluster B disorder, and 30 had a cluster C disorder). The distribution of patients with or without axis II comorbid disorders did not significantly differ among the three genotype groups (2 = 0.52, df = 2, p = NS). Forty (32%) patients had concomitant axis I and II disorders; the distribution of patients with or without concomitant axis I and II comorbid disorders did not significantly differ among the three genotype groups (2 = 0.43, df = 2, p = NS). These results did not change when SS and SL subjects were combined together and analyzed against LL patients.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Consistently with the present results, a lack of association between 5HTTLPR genotype and BN has been reported by two research groups (19,20), but not by Di Bella et al (18) who found a higher prevalence of the S allele in patients with BN and reported an increased risk for BN in subjects carrying at least one copy of this allele. However, it must be pointed out that in the Di Bella et al study (18), only 50 bulimics were investigated, and the 5HTTLPR genotype distribution in patients deranged from the Hardy-Weinberg equilibrium. Furthermore, we found a higher frequency of the L allele in bulimic individuals. Although these differences could be related to the nature of the control groups that may be responsible for variations in allele frequencies occurring among the studies, it is worth mentioning that Matsushita et al (29) found a significantly higher frequency of the L allele in nonclinical subjects scoring abnormally high on the Eating Attitude Test. However, because the S form of the 5HTTLPR is associated with a lower 5HTT transcriptional activity (17), the findings of a reduced hypothalamic and thalamic 5HTT availability in bulimic patients (14) and of a decreased platelet 3[H]paroxetine binding in individuals with bulimic syndromes carrying the S allele (30) would support an increased frequency of this allele in BN. Furthermore, it has been recently suggested that the S allele could represent a moderate but significant risk factor for anorexia nervosa (31). Therefore, at present, the issue of the 5HTTLPR allele frequency in eating disorders and, in particular, in BN is far from being resolved.
Interestingly, differently from Steiger et al (30) who did not report any significant association between 5HTTLPR and BMI in individuals with bulimic syndromes, our patients with BN with the SS or SL genotype had reduced BMI values as compared with LL individuals, this effect being more pronounced in homozygote SS patients and essentially as a result of a reduced body fat mass. Differences in the size and clinical composition of the patient samples could explain this discrepancy, because in the Steiger et al study, the patient sample consisted of only 59 women, including patients with either BN or binge/purge anorexia nervosa or eating disorders not otherwise specified. Because the S allele is associated with a lower transcriptional activity of the 5HTT gene and a reduced 5HT reuptake efficiency (17), it can be tentatively argued that patients with SS or SL genotypes have an enhanced availability of 5HT at their central serotonergic synapses, which, in turn, potentiate satiety and reduce food intake (1) favoring a lower BMI and a decreased fat mass. It is evident from the present findings that the putative enhanced serotonergic transmission in SS and SL patients cannot be the sole determinant of decreased body fat stores, because this phenomenon did not occur in our healthy control subjects. Because the S allele resulted associated with both decreased BMI/body fat mass and increased harm avoidance scores in patients with BN, the possibility existed that higher harm avoidance was mediating the association between this allele and the anthropometric variables. Correlation analyses showed that this was not the case. Other factors, not explored in this study and likely interacting with genotype, were probably involved in the reduced BMI phenotype of patients with BN carrying at least one copy of the S allele. Therefore, according to the arguments recently made by Moffitt et al (32), this would suggest a form of gene by environment interaction such that the gene effect is present only in women with a certain exposure.
The major finding of this study was that women with BN exhibited higher levels of harm avoidance as compared with control women and that carriers of the 5HTTLPR S allele (SS and SL genotypes) exhibited higher harm avoidance scores than noncarriers. In line with our findings, previous studies of comparison between bulimic patients and healthy control subjects have consistently reported the occurrence of increased harm avoidance in women with BN (7–9). All these studies used the nonrevised version of the Cloninger’s TCI, which is a 240-item instrument designed to measure personality dimensions using a true/false questionnaire. The TCI-R uses a five-point scale (definitely false, mostly or probably false, neither true nor false, or about equally true or false, mostly or probably true, definitely true) to score the 240 items. Therefore, present results based on the TCI-R give further support to the idea that increased harm avoidance is associated with active states of BN.
It is widely accepted that human personality traits are genetically influenced to a considerable degree; up to 40% to 60% might be explained by genetic factors (33). Cloninger (6) hypothesized that individual differences in harm avoidance are mediated by genetic variability in 5HT transmission. A recent meta-analysis (34) have reported that at least 26 studies investigated the putative association between 5HTTLPR and anxiety-related personality traits, including harm avoidance. The results of these studies have been inconsistent with some studies finding evidence for an association and others not. This inconsistency has been explained by the heterogeneity of the studied populations in terms of sizes, ethnic and sociocultural characteristics, and especially by differing methods of personality assessment. Anyway, to the best of our knowledge, no study has assessed the influence of 5HTTLPR genotype on harm avoidance in patients with BN. Therefore, the present findings are the first showing that in a sample of purging bulimic women, the biallelic functional polymorphism of the human 5HTT gene regulatory region contributes significantly to variation in the harm avoidance personality trait. This association was not present in our healthy control subjects, supporting the increasingly accepted view that differential gene effects may be operative in distinct normal and clinical populations. Of note, genetic influences are not the only factors that lead to individual differences in personality dimensions, behavior, and psychopathology. Complex traits are most likely to be generated by a complex interaction of environmental and experiential factors with a number of genes and their products.
Finally, comparison of the 5HTTLPR distribution in the presence or absence of comorbid psychiatric disorders revealed that there were no significant associations between the 5HTTLPR genotype and both axis I and II comorbid psychiatric disorders in women with purging BN. Similarly, no significant association emerged between 5HTTLPR genotype and a history of anorexia nervosa in our patient sample. According to our findings, Matsushita et al (20) reported no significant association between the 5HTTLPR genotype and both axis I and II comorbid psychiatric disorders in a sample of Japanese women with anorexia nervosa or BN. Therefore, it seems that 5HTTLPR genotype is not a major determinant of psychiatric comorbidity in patients with eating disorders.
In conclusion, in our sample of normal-eating control subjects and purging-type bulimics, no difference was found in the distribution of 5HTTLPR genotypes, although the L allele was more frequent in the latter; moreover, the S form of the 5HTTLPR was significantly associated in women with BN with reduced BMI and body mass values and increased harm avoidance score. These findings seem to support the view that polymorphic variants of the 5HTT promoter region do not play a part in predisposing to BN, whereas they seem to predispose bulimic individuals to nutritional impairment and increased harm avoidance.
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Supported by grant n. MM06493915_001 of the Italian Ministry of University and Scientific Research.
DOI:10.1097/01.psy.0000195746.52074.63
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  A. Favaro, E. Tenconi, and P. Santonastaso The Relationship Between Obstetric Complications and Temperament in Eating Disorders: A Mediation Hypothesis Psychosom Med, April 1, 2008; 70(3): 372 - 377. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
