Reduced 5-HT2A Receptor Binding After Recovery from Anorexia Nervosa

Reduced 5-HT2A Receptor Binding after Recovery from Anorexia Nervosa

Guido K. Frank, Walter H. Kaye, Carolyn C. Meltzer, Julie C. Price, Phil Greer, Claire McConaha, and Kelli Skovira

Background: Several lines of evidence suggest that a thinness, obsessive fears of being fat, and emaciation. disturbance of serotonin neuronal pathways may contrib- Consequently, AN has the highest mortality among the ute to the pathogenesis of anorexia nervosa (AN). This psychiatric disorders (Sullivan 1995). Depression, anxiety, study applied positron emission tomography (PET) to and obsessive–compulsive behaviors are common. People investigate the brain serotonin 2A (5HT2A) receptor, with AN tend to have a characteristic cluster of personality which could contribute to disturbances of appetite and and temperament traits including perfectionism, over con- behavior in AN. trol, rigidity, and harm avoidance (Palmer and Jones Methods: To avoid the confounding effects of malnutri- 1939). Such behaviors may be traits because they occur tion, we studied 16 women recovered from AN (REC AN, premorbidly and persist after weight and eating normalize Ͼ 1 year normal weight, regular menstrual cycles, no (Casper 1990; Klump et al 2000; Srinivasagam et al 1995). bingeing or purging) compared with 23 healthy control The etiology of AN is presumed to be complex and has women (CW) using [18F]altanserin, a specific 5-HT2A receptor antagonist on PET imaging. traditionally been thought to be influenced by develop- mental and social factors (Garner 1993; Treasure and Results: REC AN women had significantly reduced Campbell 1994); however, the stereotypic clinical presen- [18F]altanserin binding relative to CW in mesial temporal tation, gender distribution, and age of onset suggest a (amygdala and hippocampus), as well as cingulate cortical regions. In a subset of subjects (11 CW and 16 REC biologic vulnerability. Recent studies show that AN is AN), statistical parametric mapping (SPM) confirmed familial (Lilenfeld et al 1998; Strober et al 2000) with reduced mesial temporal cortex 5HT2A receptor binding three large-scale, community-based twin studies suggest- and, in addition, showed reduced occipital and parietal ing that eating disorders have a heritability of between 0.5 cortex binding. and 0.8 (Bulik et al 1998). The clustering of the disorder Conclusions: This study extends research suggesting that in families may be related in part to a genetic transmission altered 5-HT neuronal system activity persists after recov- of risk. ery from AN and may be related to disturbances of mesial A considerable number of animal and human studies temporal lobe function. Altered 5-HT neurotransmission suggest that altered central nervous system serotonin after recovery also supports the possibility that this may (5-HT) activity may contribute to the appetitive and be a trait-related disturbance that contributes to the behavioral alterations found in AN. For example, in- pathophysiology of AN. Biol Psychiatry 2002;52: creased intrasynaptic 5-HT, or directly activated 5-HT 896–906 © 2002 Society of Biological Psychiatry receptors, tend to reduce food consumption (Blundell 1984; Leibowitz and Shor-Posner 1986), and alterations in Key Words: Anorexia nervosa, serotonin, receptor, limbic 5-HT activity be implicated in disturbances of mood and system impulse control (Higley and Linnoila 1997; Kaye 1997; Introduction Lucki 1998; Mann 1999). In fact, most studies have shown that ill AN patients have disturbances (i.e., a reduction) of norexia nervosa (AN) is a disorder that most often central 5-HT activity (Kaye et al 1988; Walsh et al 1998; Ahas its onset during adolescence in young women and Wolfe et al 1997). is invariably characterized by the relentless pursuit of Behavioral or 5-HT disturbances in ill AN patients may be a consequence of malnutrition or premorbid traits that contribute to a vulnerability to develop AN. Determining From the Department of Psychiatry, Western Psychiatric Institute and Clinic (GKF, WHK, CM, KS), and Department of Radiology, Presbyterian University whether abnormalities are a consequence or a potential Hospital (CCM, JCP, PG), University of Pittsburgh, School of Medicine, Pittsburgh, Pennsylvania. antecedent of pathologic feeding behavior is a major Address reprint requests to Walter H. Kaye, M.D., University of Pittsburgh, question in the study of eating disorders. It is impractical Western Psychiatric Institute and Clinic, 3811 O’Hara Street, Pittsburgh PA 15213. to study AN prospectively because of the young age of Received November 7, 2001; revised February 15, 2002; accepted February 21, 2002. onset and difficulty in premorbid identification of people

© 2002 Society of Biological Psychiatry 0006-3223/02/$22.00 PII S0006-3223(02)01378-1 Serotonin Receptor Activity in Anorexia BIOL PSYCHIATRY 897 2002;52:896–906 who will develop the disorder. Another strategy is to study disorders treatment program at the Western Psychiatric Institute women who are recovered from AN. Any persistent and Clinic (Pittsburgh, PA) or were recruited through advertise- psychobiological abnormalities might be trait related and ments. All subjects underwent four levels of screening: 1) a brief potentially contribute to the pathogenesis of this disorder. phone screening; 2) an intensive screening assessing psychiatric In fact, increased levels of cerebrospinal fluid (CSF) history, lifetime weight, and exercise and menstrual cycle history 5-hydroxy indoleacetic acid (5-HIAA), the major metab- as well as eating pattern and exercise history for the past 12 olite of 5-HT, have been found after long-term recovery months; 3) a comprehensive assessment using structured and semistructured interviews; and 4) a face-to-face interview with a from AN (Kaye et al 1991). Other studies of recovered AN psychiatrist. To be considered “recovered,” subjects had to 1) found reduced appetite suppression in response to the maintain a weight above 85% average body weight (Metropoli- 5-HT challenge drug D-fenfluramine (Ward et al 1998), tan Life Insurance Company 1959); 2) have regular menstrual but normal 5-HT-related endocrine secretion (O’Dwyer et cycles; and 3) have not binged, purged, or engaged in significant al 1996; Ward et al 1998). restrictive eating patterns for at least 1 year before the study. The 5-HT neurotransmitter system is widespread Additionally, subjects must not have used psychoactive medica- throughout the brain, (Nieuwenhuys et al 1988), and the tion such as antidepressants or met criteria for alcohol or drug loci of CNS pathophysiology responsible for serotonin and abuse or dependence, major depressive disorder or severe anxiety behavioral abnormalities in people with AN are unknown. disorder within 3 months of the study. Twenty-one healthy Positron emission tomography (PET) technologies hold control women (CW) were recruited through local advertise- the promise of understanding previously inaccessible brain ments. The CW had no history of an eating disorder or any 5-HT function in vivo and its dynamic relationship with psychiatric, medical, or neurologic illness. They had no first- human behaviors. In this study, the radioligand [18F]al- degree relative with an eating disorder. They had normal men- tanserin, a specific 5-HT2A receptor antagonist, was used strual cycles and had been within normal weight range since together with PET imaging to determine whether the menarche. 5-HT2A receptor system was altered in women who were This study was conducted according to local institutional review board regulations, and all subjects gave written informed recovered from AN. The 5-HT2A receptor system is of consent. The PET imaging was performed during the first 10 potential interest in AN because it has been implicated in days of the follicular phase for all subjects. Subjects were the modulation of feeding (inducing hypophagia), mood, admitted to a research laboratory on the eating disorders unit of and anxiety as well as in antidepressant efficacy (Bon- Western Psychiatric Institute and Clinic at 9:00 pm of the day homme and Esposito 1998; De Vry and Schreiber 2000; before the PET study for adaptation to the laboratory and for Simansky 1996; Stockmeier 1997). In addition, some but psychologic assessments. The PET study was done the next day. not all studies have found differences in allele frequencies All subjects had the same standardized, monoamine controlled of the 5-HT2A receptor in AN (Hinney et al 2000). (low protein) breakfast on the morning of the study. Because 5-HT2A receptors are in high concentrations in Blood was drawn for assessment of ␤-hydroxybutyrate the cortex, this study sought to determine whether PET (BHBA), a plasma ketone body that is relatively sensitive to could identify alterations of specific cortical brain regions reflecting the presence of starvation (Fichter et al 1990), as well in AN. as for evaluation of gonadal hormone levels (estradiol, E2). The Recently, we showed in a group of women recovered Structured Clinical Interview for DSM-IV Axis I Disorders (First from bulimia nervosa (REC BN) reduced cortical 5-HT2A et al 1996) was used to assess the lifetime prevalence of Axis I receptor binding that was significant in the orbital frontal psychiatric disorders, and the Structured Interview for Anorexia and Bulimia (Fichter et al 1998) to assess lifetime diagnosis of an cortex (Kaye et al 2001). Women with AN and BN share eating disorder. These interviews were administered by a trained increased CSF 5-HIAA after recovery (Kaye et al 1998a, masters- or doctoral-level clinical interviewer and blindly re- 1998b). Thus, we hypothesized that postsynaptic cortical viewed by a psychiatrist or trained PhD-level psychologist, who 5-HT2A receptor densities would be downregulated in validated final DSM-IV Axis I diagnoses. Current psychopathol- response to the suggested 5-HT hyperactivity in AN after ogy was assessed with a battery of standardized instruments recovery (Rioux et al 1999; Roth et al 1998) similar to our including the Beck Depression Inventory (Beck et al 1961), the previous finding in REC BN. We hypothesized that, if Spielberger State–Trait Anxiety Inventory (Spielberger et al regional cortical alterations occur in AN, they are most 1970), the Frost Multidimensional Perfectionism Scale (Frost et likely to be present in frontal or limbic regions because al 1990), and the Temperament and Character Inventory (Clon- such regions have been particularly implicated in the modu- inger et al 1994) for assessment of harm avoidance, novelty lation of mood and impulse control (McAllister 1992). seeking, and reward dependence. All subjects underwent magnetic resonance (MR) imaging Methods and Materials before the PET scan on a Signa 1.5 Tesla scanner (GE Medical Systems, Milwaukee, WI) using a standard head coil. A volu- Sixteen women who had recovered from restricting type AN metric spoiled gradient recalled (SPGR) sequence (TE ϭ 5, TR (REC RAN) were recruited. No subject had a history of bingeing ϭ 25, flip angle ϭ 40°, NEX ϭ 1; field of view ϭ 24 cm, image or purging. Subjects were previously treated in the eating matrix ϭ 256 ϫ 192 pixels) acquired in the coronal plane was 898 BIOL PSYCHIATRY G.K. Frank et al 2002;52:896–906

used to guide region of interest (ROI) selection. Fast spin-echo (HPLC). Plasma data were corrected for the presence of radio- T2 and proton density weighted images were also routinely labeled metabolites of [18F]altanserin using the HPLC data acquired to exclude significant neuropathology. Pixels that cor- (Lopresti et al 1998). The PET data were also corrected for responded to scalp and calvarium were removed from the SPGR radioactive decay and scatter and reconstructed with a Hanning MR images (Sandor and Leahy 1997), and the MR and PET window and a cutoff at 0.8 of the Nyquist rate. image data were then coregistered (Woods et al 1993). The ROIs were hand drawn on the coregistered MR images The 5-HT2A receptor–specific radioligand [18F]altanserin and applied to the dynamic PET data to generate time-activity was used and synthesized as described by Lemaire et al (1991). curves. The following ROIs were selected: prefrontal cortex Subjects were studied using two PET scanners, ECAT HRϩ and (Brodmann’s area [BA] 10), medial orbital frontal cortex (BA 951R/31 (CTI PET Systems, Knoxville, TN) because the 951R- 11), lateral orbital frontal cortex (BA 47), mesial temporal cortex PET scanner was replaced by the HRϩ during the course of this (amygdalo–hippocampal complex), lateral temporal cortex (BA study. The 951R/31 acquires 31 slices over a 108-mm axial field 21), supragenual cingulate (BA 24/32, five planes superior to of view (FOV). The HRϩ scanner acquires 63 continuous slices anterior most part of genu corporis callosi), pregenual cingulate over a 152-mm axial FOV. All subjects were studied in two (BA 24/32, anterior to anterior most part of genu of the corpus dimensional (2D) imaging mode (septa extended). Under these callosum), and subgenual cingulate (BA 25, inferior to the genu conditions, the 951R/31 has a point source resolution in air of the corpus callosum), parietal cortex (BA 7), and sensorimotor reconstructed (without smoothing) of approximately 6.0 mm cortex (BA 1,2,3,4). We also performed ROI sampling of the full-width at half-maximum (FWHM) in the in-plane, and 5.0 cerebellum, and this was used as the reference region because of mm FWHM in the axial orientation (Spinks et al 1992). In 2D, the low concentration of 5-HT2A receptors (Pazos et al 1987). ϩ the HR scanner has an approximated in-plane resolution of 4.9 The ROIs were expressed as the mean of left and right values. All mm (1–10 cm radius) at FWHM, and an axial resolution of 4.4 subjects were assigned a number at the local PET facility, and mm FWHM (Brix et al 1997). All REC AN women were studied imaging data were analyzed blindly. The data on 12 CW from the ϩ ϩ on the HR PET scanner. Eleven CW were studied on the HR 951R scanner had previously been assessed similarly under blind scanner and 12 were studied on the 951R PET scanner. Data on conditions (Kaye et al 2001). For the kinetic analyses, the Logan the 12 CW studied with the 951R PET scanner were previously graphic method was applied based on sampled ROI data from 8 published (Kaye et al 2001). to 90 min following [18F]altanserin injection. The regression The CW studies on the HRϩ and the 951R PET scanners had slope value (distribution volume, DV) for each ROI was calcu- similar age and body mass index (BMI). A comparison of lated (Logan et al 1990). The cerebellar data were assumed to regional 5-HT2A receptor binding potential values between CW represent free and nonspecifically bound radioactivity concen- sub groups using nonparametric Mann–Whitney U tests yielded trations. The ratio of regional [18F]altanserin binding (DV ROI) similar binding across groups, with a U value range between 5 in relation to cerebellar [18F]altanserin binding (DV Cer) was and 52, and corresponding p values between 0.9 and 0.2. For calculated (DV /DV ϭ Dvratio, DVR), and the DVR was example, CW values were as follows: prefrontal cortex, 2.41 Ϯ ROI Cer used as a specific binding measure (Smith et al 1998). Recent 0.2 (HRϩ scanner) and 2.40 Ϯ 0.3 (951R scanner), U ϭ 50, p ϭ literature suggests that the cerebellum is not totally devoid of .8; mesial temporal cortex, 1.64 Ϯ 0.2 (HRϩ scanner) and 1.70 Ϯ 0.2 (951R scanner), U ϭ 46, p ϭ .6. Because these two groups 5HT2A receptors (Dwivedi and Pandey 1998; Staley et al 2001). of CW were similar, and because the methods for 5HT2A Although the number of such cerebellar 5HT2A receptors is low, binding assessment used relative measures (regional binding an influence on ROI-specific binding could not be excluded. We divided by the binding in a reference region, discussed below), therefore also compared the cerebellar DV between groups. the data on CW were combined. The ROI-based data analysis may be confounded by how Subjects were positioned with the head oriented parallel to the individual regions are drawn, as well as by potentially missing cantomeatal line. A softened thermoplastic mold with generous areas of binding alterations due to a limited number of assessed holes for eyes, nose, and ears was fitted closely around the head ROIs. Thus, an additional analysis using statistical parametric ϩ and attached to the headholder to minimize subject motion. A mapping was performed on all subjects studied on the HR 10-min transmission scan was obtained and used for attenuation scanner, using Statistical Parametric Mapping (SPM) software correction. Following bolus intravenous injection of 10 mCi (Frackowiak et al 1997; Friston 1995). We created DVR images high-specific activity (Ն1.04 Ci/␮mol) [18F]altanserin, dynamic for all subjects on the HRϩ scanner by applying the Logan emission scanning with arterial blood sampling was performed modeling method on a pixel-by-pixel basis for each PET image until 90 min postinjection. The arterial input function was and dividing each pixel by the cerebellar binding value (Smith et determined by approximately 35 0.5-mL hand-drawn samples al 1998). All MR images were normalized to the Montreal collected over the scanning interval (including 20 samples in the Neurologic Institute (MNI, Canada) standard brain that is derived initial 2 min postinjection). Blood samples were centrifuged and from 152 subjects (Evans et al 1993). The normalized DVR the plasma radioactivity concentration measured (Cobra II, images were then smoothed with a Gaussian filter (12 mm Packard Instruments, Cleveland, OH). Additionally, 3-mL blood FWHM). Statistical parametric maps were created based on the samples were acquired at 2, 10, 30, 60, and 90 min after t statistic (Friston 1995), assigning a p value to each voxel [18F]altanserin injection and used to determine the fraction of (probability image). An initial uncorrected p value of Ͻ0.05 was unmetabolized [18F]altanserin (of total plasma radioactivity selected as threshold for display in the probability image. concentration) using high-performance liquid chromatography Regions considered significant had to be either of significant Serotonin Receptor Activity in Anorexia BIOL PSYCHIATRY 899 2002;52:896–906

Table 1. Demographic Variables between Groups and Behavioral Assessment Data

CW (n ϭ 23) REC AN (n ϭ 16) Mean SD Mean SD U Exact Sig. Age (years) 25.3 5.3 24.5 6.1 155 .4 Current BMI 21.4 1.3 20.5 1.8 105 .024 AN onset (years of age) ——14.2 2.9 —— Duration of recovery (years) —— 4.3 3.3 —— Estradiol (pmol/mL) 35 26 74 51 28.5 .1 Plasma BHBA (␮mol/mL) 100 103 42 14 31 .1 Drive for Thinness (EDI) 0.6 1.2 15.3 (12) 6.4 14.5 Ͻ.001 Depression (BDI) 1 1.5 3.3 (12) 3.6 87.5 .047 Perfectionism (MPS) 54 13 101 (12) 29 28.5 Ͻ.001 Novelty seeking (TCI) 21 5 20 (12) 7 109 .5 Harm avoidance (TCI) 12 4 15 (12) 8 87 .15 Reward dependence (TCI) 20 2 16 (12) 4 59 .01 State anxiety (STAI) 26 5 32 (12) 7 78 .04 Trait anxiety (STAI) 28 7 38 (12) 10 53.5 .04

(12) indicates that only 12 of the REC AN women completed self-assessment questionnaires. Group comparison by Mann–Whitney U test. CW, healthy comparison women; REC AN, recovered anorexic women; Sig., significance; BMI, body mass index; BHBA, beta hydroxy butryic acid; EDI, Eating Disorders Inventory; TCI, Temperament and Character Inventory; BDI, Beck Depression Inventory; STAI, State and Trait Anxiety Inventory.

cluster size or voxel significance (p Ͻ .05) after correction for Results multiple comparisons. We also assessed, in an exploratory analysis, possible struc- Demographic Variables and Behavioral tural differences between groups, because such alterations theo- Assessments retically could confine 5-HT receptor distribution findings. We The REC AN and CW women were of similar ages (Table used SPGR images and compared these by group with the 1). Although current BMI was significantly lower in REC voxel-based morphometry tool kit (VBM) within SPM (Ash- AN, the mean difference was less than 1 BMI unit. Subject burner and Friston 2000). In short, this analysis is a voxelwise groups had similar plasma BHBA values, a measure of comparison of the local concentration of gray and white matter. For this approach, SPGR images from each subject were aligned ketone body metabolism, suggesting REC AN were not to a vertical, midsagittal plane (Woods et al 1992) and spatially starving. In addition, plasma E2 was similar between groups. transformed into the MNI standard reference atlas described The REC AN subjects scored significantly higher on above. All images were separated into GM, WM, and CSF. drive for thinness, perfectionism, state and trait anxiety, Differences in between-subject global activity were removed by and depressive symptoms (Table 1). Groups did not differ analysis of covariance (ANCOVA) on a pixel-by-pixel basis with significantly on harm avoidance or novelty seeking; how- global counts as the covariate. The resulting image data were ever, reward dependence was significantly reduced in then convolved with a three-dimensional Gaussian filter (12 mm REC AN. Eight subjects of the REC AN group had a FWHM). Groups were compared using t statistic–based para- confirmed history of major depressive disorder (MDD), metric maps and multiple comparison corrections as above. three a subthreshold MDD, and seven subjects had a The SPSS statistical software package (Barcikowski 1984) history of obsessive–compulsive disorder (OCD). One was used for all other statistical analyses. Because of the subject in the REC AN group had a history of a panic relatively small sample size, between-group comparisons were made with nonparametric Mann–Whitney U two independent disorder. None of the REC AN women had a history of samples tests calculating two-sided exact significance levels. any psychotic disorder. Multiple comparisons were corrected using the Bonferroni correction. Correlations were examined with Spearman correlation ROI-Based Analysis of 5-HT2A Receptor Binding coefficients. Factors that were considered possible confounds of 5-HT2A receptor binding were explored with analyses of covari- The REC AN subjects had a significant reduction of ance (ANCOVA). To assess possible seasonal influences on 5-HT2A receptor binding in mesial temporal cortex, pre- 5HT2A binding, we compared the distribution of PET scans over and subgenual cingulate (Table 2) and a trend toward a the 12 months between groups using a chi-square test. All values reduction in the sensorimotor cortex. The REC AN and are expressed as mean Ϯ SD. As level of significance a p value CW women had similar binding in the prefrontal cortex, of p Ͻ .05 was selected. medial, and lateral orbital frontal cortex and the lateral 900 BIOL PSYCHIATRY G.K. Frank et al 2002;52:896–906

Table 2. Regional 5-HT2A Receptor Binding between Groups

CW (n ϭ 23) REC AN (n ϭ 16)

DVROI/DVCer DVROI/DVCer Region of interest Mean SD Mean SD U Exact Sig. Prefrontal cortex 2.42 0.25 2.41 0.26 183 .9 Lateral orbital frontal cortex 2.39 0.25 2.4 0.26 170 .7 Medial orbital frontal cortex 2.58 0.28 2.51 0.32 155 .4 Supragenual cingulate 2.45 0.31 2.29 0.25 129 .1 Subgenual cingulate 2.69 0.27 2.47 0.27 110 .035 Pregenual cingulate 2.58 0.28 2.4 0.24 113 .04 Lateral temporal cortex 2.63 0.22 2.65 0.29 179 .9 Mesial temporal cortex 1.68 0.19 1.52 0.17 86 .004 Parietal cortex 2.6 0.28 2.49 0.32 127 .2 Sensorimotor cortex 2.28 0.28 2.13 0.27 124 .09

Group comparison by Mann–Whitney U test. DVROI/DVCER is the distribution volume of regional serotonin 2A (5-HT2A) receptor binding in relation to cerebellar value. CW, healthy comparison women; REC AN, recovered anorexic women; Sig., significance.

temporal cortex. Ten ROIs were assessed. After correction 30, 60, and 90 min) using a repeated measures ANOVA. for multiple comparisons, REC AN women continued to Data between CW and REC AN were similar (F ϭ 0.6, have reduced binding in the mesial temporal cortex (p ϭ p ϭ .7). .04). Because the CW women were studied on two scanners, a comparison was done between CW (n ϭ 11) and REC SPM-Based Analysis of 5-HT2A Receptor Binding AN (n ϭ 16) women who were scanned on the same This analysis revealed significant clusters of reduced scanner (ECAT HRϩ). Both groups were of similar ages 5-HT2A receptor binding in REC AN compared with CW and BMI. The REC AN subjects continued to show after correction for multiple comparisons. In the left significantly reduced binding in the mesial temporal cor- hemisphere, these regions were located in the mesial tex (p ϭ .008) and anterior medial cingulate (p ϭ .02) and temporal cortex, middle temporal gyrus, and subcortical a trend toward reduced binding in the supragenual (p ϭ parietal lobe (Table 3 and Figure 1). In the right hemi- .08) and subgenual (p ϭ .05) cingulate, as well as the sphere, these regions were located in the middle temporal lateral orbital frontal (p ϭ .1), parietal (p ϭ .06), and gyrus, amygdaloid complex, and subiculum–hippocampus. sensorimotor cortex (p ϭ .09). The cerebellar 5HT2A DV between CW (1.36 Ϯ 0.18) and REC AN (1.23 Ϯ 0.37) was similar (U ϭ 118, p ϭ .1). Analysis of Structural Images Using VBM/SPM In addition, we compared the percentage of unmetabolized There was no significant difference between groups on a [18F]altanserin (parent) in plasma between control and voxel level. In particular, no mesial temporal or cingulate anorexic subjects five times throughout the study (2, 10, related alterations were found.

Table 3. Cluster Sizes, p values, and Z Scores for Areas of Most Prominent 5-HT2A Binding Alterations for Right (r) and Left (l) Hemispheres for Statistical Parametric Mapping–Based Analysis

Cluster Coordinates (mm) Size p (corrected)Z score x y z Brain Region 4478 .001 4.31 Ϫ50 Ϫ44 Ϫ14 Middle temporal gyrus (l) 3.80 Ϫ30 Ϫ28 Ϫ22 Mesial temporal cortex (l) 3.43 Ϫ38 Ϫ52 26 Subcortical parietal lobe (l) 7446 Ͻ.001 3.77 32 4 Ϫ36 Amygdaloid complex (r) 3.42 26 Ϫ12 Ϫ30 Subiculum/hippocamps (r) 3.25 56 Ϫ40 Ϫ16 Middle temporal gyrus (r)

Subjects scanned on the HRϩ scanner (CW n ϭ 11; REC AN n ϭ 16). CW, control women; REC AN, recovered anorexic women. Serotonin Receptor Activity in Anorexia BIOL PSYCHIATRY 901 2002;52:896–906

reduced 5HT2A receptor binding compared with CW for the mesial temporal cortex (1.4 Ϯ 0.05 vs. 1.7 Ϯ 0.2, p ϭ .001; p ϭ .01 after correction for multiple comparisons) and a tendency toward reduced binding (p Ͻ .1) in sensorimotor, parietal, subgenual cingulate cortex, and basal ganglia (data not shown). Neither the CW nor the REC AN women had significant relationships between measures of behavior and 5-HT2A binding. There was no difference between groups in terms of the relationship between 5HT2A receptor binding and seasonality (X ϭ 2.5, df ϭ 1, p ϭ .2).

Discussion Figure 1. Clusters of significantly reduced 5HT2A binding in These data confirm that a disturbance of brain 5-HT recovered anorexic women (REC AN, n ϭ 16) compared with neuronal function persists after recovery from AN. Di- healthy comparison women (CW, n ϭ 11) with all subjects scanned on the HRϩ scanner. p Ͻ .05 for each voxel between rected ROI and whole brain SPM analyses suggested that groups, df ϭ 1,25 based on the t-statistic; all clusters shown are the most substantial reduction in 5-HT2A receptor binding significant after correction for multiple comparisons (p Յ .001). occurred in the mesial temporal cortex, including the The upper image row shows significant clusters in a standard amygdala and hippocampus. grid system. The lower row overlays significant clusters on a Recovered AN women have increased CSF 5-HIAA normalized reference MRI. raising the possibility that they have increased brain 5-HT activity (Kaye et al 1991). One possible explanation for Relationships of Demographic Values and the findings in this study is that reduced 5-HT2A binding Comorbidity with 5HT2A Receptor Binding is a compensatory response for increased extracellular levels of 5-HT. Studies in animals confirm that reduced The CW and REC AN groups each showed significantly 5-HT2A receptor density occurs in response to increased negative relationships between age and 5-HT2A receptor intra synaptic 5-HT (Rioux et al 1999; Saucier et al 1998) binding for most cortical regions (Table 4). A comparison or 5HT agonists (Eison and Mullins 1996). In addition, of slopes shows similarities for all regions for REC AN paroxetine treatment for depression reduces 5-HT2A re- and CW. Neither group had significant relationships be- ceptors in the brain, presumably due to increased intrasyn- tween 5-HT2A receptor binding and current BMI, plasma aptic 5-HT (Meyer et al 2001). Some, but not all, studies BHBA, or E2. Subgroups of REC AN subjects had similar raise the question of whether an increased frequency of a regional binding when compared as to whether they had a variant of the 2A receptor is present in AN (Hinney et al lifetime history of MDD or OCD. The REC AN without a 2000); however, it is unknown if this receptor variant has ϭ history of MDD (n 5) continued to show significantly functional significance. It is also important to note that these PET imaging studies do not permit a determination Table 4. Regional 5-HT2A Receptor Binding and Correlation of whether there is an alteration in affinity or number of with Age the 5-HT2A receptor. CW REC AN We are not aware of any previous in vivo neuroreceptor (n ϭ 23) (n ϭ 16) studies in ill or recovered AN women; however, 5-HT2, or rho p rho p more specifically 5-HT2A receptor binding, has been assessed in depressed subjects. Some studies of MDD Ϫ Ϫ Prefrontal cortex .5 .01 .4 .13 have found reduced 5-HT2A receptor binding in frontal, Lateral orbital frontal cortex Ϫ.3 .23 Ϫ.6 .01 Medial orbital frontal cortex Ϫ.5 .03 Ϫ.6 .01 temporal, parietal, and occipital cortical regions before or Supragenual cingulate Ϫ.5 .01 Ϫ.6 .02 after treatment (Attar-Levy et al 1999; Biver et al 1997; Subgenual cingulate Ϫ.5 .03 Ϫ.4 .10 Yatham et al 2000). Other studies in depressed subjects Pregenual cingulate Ϫ.5 .02 Ϫ.6 .02 (Meltzer and Reynolds 1999; Meyer et al 1999) did not Ϫ Ϫ Lateral temporal cortex .5 .03 .5 .04 find such changes. To our knowledge, no such studies Mesial temporal cortex Ϫ.3 .20 Ϫ.3 .35 Parietal cortex Ϫ.5 .01 Ϫ.5 .56 have been done in anxiety disorders or OCD. Sensorimotor cortex Ϫ.5 .01 Ϫ.6 .02 Other types of brain imaging studies in AN have

CW, healthy control women; REC AN, recovered anorexic women; rho, reported structural or functional alterations. Subjects with Spearman correlation coefficient. active AN have enlarged ventricles and sulci widening 902 BIOL PSYCHIATRY G.K. Frank et al 2002;52:896–906

(see review Ellison and Fong 1998). A 1H-MR scan This study raises the possibility that REC AN women revealed reduced lipid signals in the frontal white matter have altered 5HT2A binding in cingulate regions as well and occipital gray matter and was associated with de- as occipitoparietal and sensorimotor cortex. The anterior creased BMI (Roser et al 1999). These alterations have cingulate receives afferents from the amygdala and has been thought to be reversible after recovery but recent data direct projections to the premotor frontal cortex and other raise the question of persistent changes after recovery limbic regions. The anterior cingulate has been character- (Katzman et al 1997; Lambe et al 1997). A number of ized as “executive” in function and contributes to attention studies using single photon computed tomography by monitoring the motivational significance of stimuli (SPECT) or PET with 2-deoxy-glucose, have shown (Devinsky et al 1995; Tucker et al 1995). Overall, the temporal alterations and less frequently frontal or cingu- anterior cingulate cortex has a role in affect as well as late changes. Gordon et al (1997) using SPECT found ill visuospatial and memory functions and appears to play a AN patients had unilateral temporal lobe hypoperfusion crucial role in initiation, motivation, and goal-directed that persisted in subjects studied after weight restoration. behaviors. The primary sensory and association cortices Ellison et al (1998) using functional MR imaging, found have input into the limbic system (Charney and Deuth that ill AN patients, when viewing pictures of high-caloric 1996), and the parietal cortex may be involved in the drinks, had increased signal changes in the left insular, circuitry of depression, anxiety (Davidson et al 1999), and cingulate gyrus, and left amygdala-hippocampus region memory consolidation (Izquierdo and Medina 1997). and increased anxiety. Rauch et al (2000) using PET O-15 In another study, our group reported (Kaye et al 2001) and pictures of high-calorie food, found ill AN patients that nine REC BN women, who never had AN, had a had elevated regional cerebral blood flow in bilateral significant downregulation of 5-HT2A receptors in the mesial temporal lobes and increased anxiety. Taken to- orbital frontal cortex and a trend toward a reduction in the gether, these studies suggest that temporal cortex alter- lateral orbital frontal cortex and the sensorimotor cortex. ations occur in AN. In addition, the CW showed a well-known negative Similarly, our study found that the most pronounced relationship between age and 5-HT2A binding in most change in 5-HT2A binding in REC AN women was in the cortical regions (data on 12 of the CW in this study were mesial temporal cortex regions, including hippocampus reported in our previous study). Importantly, a relationship and amygdala regions. The amygdala is thought to play a between age and 5HT2A binding was not found in the central role in the “interpretation of fear” (Charney and REC BN sample in regions of interest. Although AN and Deuth 1996; Davis 1992) and anxiety disorders (Benkelfat BN are related disorders because they are cross-transmit- et al 1995; Reiman et al 1984; Van der Linden et al 2000). ted in families, a number of factors distinguish the The mesial temporal cortex may also play a role in subgroups, such as extremes of eating behavior and depressive symptoms (Byrum et al 1999; Kennedy et al impulse control. These data raise the possibility that 1997), OCD (Breiter and Rauch 1996; Zald et al 1996), regional differences in [18F]altanserin binding or relation- memory processing (Marrazzi et al 1990), and aversive ships of [18F]altanserin binding to age may distinguish conditioning (Breiter and Rosen 1999; Buchel et al 1999; eating disorder subgroups after recovery. That is, REC BN Schneider et al 1999; Zalla et al 2000). Thus, these data women had a substantial reduction of [18F]altanserin support the possibility, raised by recent neuroimaging 5HT2A receptor binding localized to the orbital frontal studies, that alterations of the hippocampus–amygdala cortex (Kaye et al 2001). In contrast, REC AN show region may play a more general role in psychopathology diminished [18F]altanserin in temporal as well as cingu- (Fujita et al 2000). In a preliminary analysis, 5HT2A late, parietal, and sensorimotor binding in this study. The binding changes in our study were generally bilateral and disturbance of orbital frontal 5HT circuits in BN may not lateralized (data not shown); however, due to the contribute to a vulnerability for imprecise and poorly relatively small sample size and increased variability when modulated self-control, whereas temporocingulate alter- assessing lateralized binding, we cannot exclude a lataral- ations in AN may contribute to impaired motivation and ization of altered binding in a larger group of subjects. integration of cognition and mood. Consistent with earlier studies (Srinivasagam et al In terms of limitations, we studied a relatively small 1995), REC AN women had increased depression, anxiety, number of AN subjects; however, a power analysis using and perfectionism. In fact, such symptoms, obsessions the data for the mesial temporal cortex showed a power of (Kaye 1997), and cognitive alterations (Bradley et al 1997; 0.85 for detection of two-group differences for the sample Strupp et al 1986) occur when people with AN are ill and size in this study. Still, it is possible that we did not detect persist after recovery. Theoretically, disturbed 5-HT neu- more subtle 5-HT2A receptor alterations because of the ronal transmission in the mesial temporal cortex areas in small sample size. The REC AN women had a reduction of people with AN could be related to such symptoms. current BMI compared with CW; however, all subjects’ Serotonin Receptor Activity in Anorexia BIOL PSYCHIATRY 903 2002;52:896–906 weights were in a normal weight range, and there was less studied on the HRϩ scanner only (both by ROI- and than 1 unit of BMI difference between groups. An addi- SPM-based analyses), further supports the findings in the tional regression analysis using group and BMI as regres- combined group. sors showed a significant effect for group (t ϭϪ2.5, p ϭ It is possible that structural alterations in the REC AN .02) but no effect for BMI (t ϭϪ0.1, p ϭ .9). Moreover, group might have contributed to this finding (Katzman et no previous study demonstrated a relation of BMI and al 1997), but in our investigational analysis of gray matter brain 5-HT2A receptor binding, and there was no relation volumes between groups, no structural alterations were of current BMI and receptor binding in either one of the found. groups of this study. Thus, there is no evidence that In conclusion, this study supports previous findings of differences in weight account for the findings. The use of altered 5-HT neuronal transmission after recovery from self-assessments for the evaluation of recovery may be a AN. It is problematic to identify women with AN before limitation, and different eating patterns between groups they develop the disorder. Studying women after long- could theoretically alter 5HT2A receptor activity; how- term recovery may be the best available approximation to ever, plasma BHBA was similar between groups, suggest- identifying factors that might be involved in the develop- ing a similar nutritional status. Similar plasma E2 values ment of AN. Although a scarring effects from the illness further supported normal menstrual function and status of cannot be excluded, reduced 5-HT2A receptor binding in recovery. AN thus could be an indication of a trait related 5-HT Beside [18F]altanserin, other radioligands have been disturbance or, alternatively, a secondary phenomenon in used for assessment of 5HT2A receptor imaging (Staley et response to increased central 5HT transmission in this al 1998); however, [18F]altanserin binds highly specific to group. ϭ 5HT2A receptors. It shows a high affinity (Ki 0.51 nmol/mL) and high selectivity (90- to 400-fold less potent ␣ Supported in part by National Institute of Mental Health Grant No. R01 at 1 and 5HT2C receptor) for 5HT2A receptors and is MH46001-07A1 and Children’s Hospital Clinical Research Center, superior compared with other 5HT2A ligands (Price et al Pittsburgh PA (#5 M01RR00084). 2001). The use of [18F]altanserin has been validated in the past (Smith et al 1998). A disadvantage of [18F]altanserin is its rapid metabolization that may increase nonspecific References binding. [18F]altanserin cerebellar DV values were similar Ashburner J, Friston KJ (2000): Voxel-based morphometry—the methods. Neuroimage 11:805–821. between groups in this study, however, supporting the assumption that nonspecific binding was similar between Attar-Levy D, Martinot J-L, Blin J, Dao-Castellana M-H, Crou- zel C, Mazoyer B, et al (1999): The cortical serotonin-sub-2 groups. In addition, the percentage of unmetabolized receptors studied with positron-emission tomography and [18F]altanserin in plasma was similar between control and [18F]-setoperone during depressive illness and antidepressant anorexic subjects. This suggested that altered 5HT2A treatment with clomipramine. Biol Psychiatry 45:180–186. binding between groups was not driven by different Barcikowski RS (1984): Computer Packages and Research Design, vol. 3: SPSS & SPSSX. Lanham, MD: University metabolism of [18F]altanserin. Recent studies have shown Press of America. that the cerebellum is not totally devoid of 5HT2A Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J (1961): receptors (Dwivedi and Pandey 1998; Staley et al 2001), An Inventory for measuring depression. Arch Gen Psychiatry and altered cerebellar 5HT2A binding could contribute to 4:561–571. the study outcome. Because 5HT2A cerebellar data were Benkelfat C, Bradwejn J, Meyer E, Ellenbogen M, Milot S, similar between groups, however, ROI-specific binding in Gjedde A, Evans A (1995): Functional neuroanatomy of CCK4-induced anxiety in normal healthy volunteers [see CW was not driven by reduced 5HT2A DV values in the comments]. Am J Psychiatry 152:1180–1184. reference region (i.e., cerebellum). When studies per- Biver F, Wikler D, Lotstra F, Damhaut P, Goldman S, Men- formed on different scanners are combined, different dlewicz J (1997): Serotonin 5-HT2 receptor imaging in major sensitivity or signal to noise ratio could contribute to depression: Focal changes in orbito-insular cortex. Br J altered findings; however, all PET scans were performed Psychiatry 171:444–448. in 2D mode. More important, the analysis procedure used Blundell JE (1984): Serotonin and appetite. Neuropharmacology 12B:1537–1551. was a relative measure, thus accounting for eventual systematically increased or decreased binding detection. In Bonhomme N, Esposito E (1998): Involvement of serotonin and dopamine in the mechanism of action of novel antidepressant addition, the binding values between the CW groups from drugs: A review. J Clin Psychopharmacol 18:447–454. the two scanners were similar and did not suggest any Bradley SJ, Taylor MJ, Rovet JF, Goldberg E, Hood J, Wachs- systematic errors. The comparison of groups with subjects muth R, et al (1997): Assessment of brain function in 904 BIOL PSYCHIATRY G.K. Frank et al 2002;52:896–906

adolescent anorexia nervosa before and after weight gain. Evans AC, Collins DL, Mills SR, Brown ED, Kelly RL, Peters J Clin Exp Neuropsychol 1:20–33. TM (1993): 3D statistical neuroanatomical models from 305 Breiter HC, Rauch SL (1996): Functional MRI and the study of MRI volumes. Proceedings of the IEEE-Nuclear Science OCD: From symptom provocation to cognitive-behavioral Symposium and Medical Imaging Conference, San Francisco, probes of cortico-striatal systems and the amygdala. Neuro- CA, November. image 4:S127–138. Fichter MM, Herpertz S, Quadflieg N, Herpertz-Dahlmann B Breiter HC, Rosen BR (1999): Functional magnetic resonance (1998): Structured Interview for Anorexic and Bulimic dis- imaging of brain reward circuitry in the human. Ann N Y Acad orders for DSM-IV and ICD-10: updated (third) revision. Int Sci 877:523–547. J Eat Disord 24:227–249. Brix G, Zaers J, Adam LE, Bellemann ME, Ostertag H, Trojan Fichter MM, Pirke KM, Pollinger J, Wolfram GM, Brunner E H, et al (1997): Performance evaluation of a whole-body PET (1990): Disturbances in the hypothalamo-pituitary-adrenal scanner using the NEMA protocol. National Electrical Man- and other neuroendocrine axes in bulimia. Biol Psychiatry ufacturers Association. J Nucl Med 38:1614–1623. 27:1021–1037. Buchel C, Dolan RJ, Armony JL, Friston KJ (1999): Amygdala- First MB, Gibbon M, Spitzer RL, Williams JBW (1996): Users hippocampal involvement in human aversive trace condition- Guide for the Structured Clinical Interview for DSM-IV Axis ing revealed through event-related functional magnetic reso- I Disorders—Research Version (SCID-I, version 2.0, Febru- nance imaging. J Neurosci 19:10869–10876. ary 1996 FINAL VERSION). New York: Biometrics Re- Bulik CM, Sullivan PF, Kendler KS (1998): Heritability of search Department, New York State Psychiatric Institute. binge-eating and broadly defined bulimia nervosa. Biol Psy- Frackowiak RSJ, Friston KJ, Frith CD, Dolan RJ, Mazziotta J, chiatry 44:1210–1218. editors (1997): Human Brain Function. San Diego, CA: Byrum CE, Ahearn EP, Krishnan KR (1999): A neuroanatomic Academic Press. model for depression. Prog Neuropsychopharmacol Biol Friston KJ (1995): Commentary and opinion: II. Statistical Psychiatry 23:175–193. parametric mapping: Ontology and current issues. J Cereb Casper RC (1990): Personality features of women with good Blood Flow Metab 15:361–370. outcome from restricting anorexia nervosa. Psychosom Med Frost RO, Marten P, Lahart C, Rosenblate R (1990): The 52:156–170. dimensions of perfectionism. Cogn Ther Res 14:449–468. Charney DS, Deutch A (1996): A functional neuroanatomy of Fujita M, Charney DS, Innis RB (2000): Imaging serotonergic anxiety and fear: Implications for the pathophysiology and neurotransmission in depression: Hippocampal pathophysiol- treatment of anxiety disorders. Crit Rev Neurobiol 10:419– ogy may mirror global brain alterations. Biol Psychiatry 446. 48:801–812. Cloninger CR, Przybeck TR, Svrakic DM, Wetzel RD (1994): The Temperament and Character Inventory (TCI): A Guide to Garner DM (1993): Pathogenesis of anorexia nervosa. Lancet its Development and Use. St. Louis, MO: Center for Psycho- 341:1631–1635. biology of Personality, Washington University. Gordon CM, Dougherty DD, Fischman AJ, Emans SJ, Grace E, Davidson RJ, Abercrombie H, Nitschke JB, Putnam K (1999): Lamm R, et al (2001): Neural substrates of anorexia nervosa: Regional brain function, emotion and disorders of emotion. A behavioral challenge study with positron emission tomog- Curr Opin Neurobiol 9:228–234. raphy. J Pediatr 139:51–57. Davis M (1992): The role of the amygdala in fear-potentiated Gordon I, Lask B, Bryant-Waugh R, Christie D, Timimi S startle: Implications for animal models of anxiety. Trend (1997): Childhood-onset anorexia nervosa: Towards identify- Pharmacol Sci 15:35–41. ing a biological substrate. Int J Eat Disord 22:159–165. Devinsky O, Morrell MJ, Vogt BA (1995): Contributions of Higley JD, Linnoila M (1997): Low central nervous system anterior cingulate cortex to behaviour. Brain 118:279–306. serotonergic activity is traitlike and correlates with impulsive De Vry J, Schreiber R (2000): Effects of selected serotonin behavior. A nonhuman primate model investigating genetic 5-HT(1) and 5-HT(2) receptor agonists on feeding behavior: and environmental influences on neurotransmission. AnnNY Possible mechanisms of action. Neurosci Biobehav Rev Acad Sci 836:39–56. 24:341–353. Hinney A, Remschmidt H, Hebebrand J (2000): Candidate gene Dwivedi Y, Pandey GN (1998): Quantitation of 5HT2A receptor polymorphisms in eating disorders. Eur J Pharmacol mRNA in human postmortem brain using competitive RT- 410:147–159. PCR. Neuroreport 9:3761–3765. Izquierdo I, Medina JH (1997): Memory formation: The se- Eison AS, Mullins UL (1996): Regulation of central 5-HT2A quence of biochemical events in the hippocampus and its receptors: A review of in vivo studies. Behav Brain Res connection to activity in other brain structures. Neurobiol 73:177–181. Learn Mem 68:285–316. Ellison AR, Foong J (1998): Neuroimaging in Eating Disorders. Katzman DK, Zipursky RB, Lambe EK, Mikulis DJ (1997): A In: Hoek HW, Treasure JL, Katzman MA, editors. Neurobi- longitudinal magnetic resonance imaging study of brain ology in the Treatment of Eating Disorders. Chichester, UK: changes in adolescents with anorexia nervosa. Arch Ped John Wiley. Adolesc Med 151:793–797. Ellison Z, Foong J, Howard R, Bullmore E, Williams S, Treasure Kaye WH (1997): Anorexia and bulimia nervosa, obsessional J (1998): Functional anatomy of calorie fear in anorexia behavior, and serotonin. In: Kaye WH, Jimerson DC, editors. nervosa [letter]. Lancet 352:1192. Eating Disorders. London: Balliere’s Tindell. Serotonin Receptor Activity in Anorexia BIOL PSYCHIATRY 905 2002;52:896–906

Kaye WH, Frank GK, Meltzer CC, Price JC, McConaha CW, Marrazzi MA, Mullings-Britton J, Stack L, Powers RJ, Lawhorn Crossan PJ (2001): Altered serotonin 2A receptor activity in J, Graham V, et al (1990): Atypical endogenous opioid women who have recovered from bulimia nervosa. Am J systems in mice in relation to an auto-addiction opioid model Psychiatry 158:1152–1155. of anorexia nervosa. Life Sci 47:1427–1435. Kaye WH, Gendall K, Strober M (1998a): Serotonin neuronal McAllister TW (1992): Neuropsychiatric sequelae of head inju- function and selective serotonin reuptake inhibitor treatment ries. Psychiatr Clin North Am 15:395–413. in anorexia and bulimia nervosa. Biol Psychiatry 44:825–838. Meltzer CC, Reynolds CF (1999): In vivo assessment of aging Kaye WH, Greeno CG, Moss H, Fernstrom J, Fernstrom M, changes in serotonin function [letter; comment]. Neuropsy- Lilenfeld LR, et al (1998b): Alterations in serotonin activity chopharmacology 21:323–324. and psychiatric symptomatology after recovery from bulimia nervosa. Arch Gen Psychiatry 55:927–935. Metropolitan Life Insurance Company (1959): New weight standards for men and women. Kaye WH, Gwirtsman HE, George DT, Ebert MH (1991): Altered serotonin activity in anorexia nervosa after long-term Meyer JH, Kapur S, Eisfeld B, Brown GM, Houle S, DaSilva J, weight restoration: Does elevated cerebrospinal fluid 5-hy- et al (2001): The effect of paroxetine on 5-HT(2A) receptors droxyindoleacetic acid level correlate with rigid and obses- in depression: an [(18)F]setoperone PET imaging study. Am J sive behavior? Arch Gen Psychiatry 48:556–562. Psychiatry 158:78–85. Kaye WH, Gwirtsman HE, George DT, Jimerson DC, Ebert MH Meyer JH, Kapur S, Houle S, DaSilva J, Owczarek B, Brown (1988): CSF 5-HIAA concentrations in anorexia nervosa: GM (1999): Prefrontal cortex 5-HT2 receptors in depression: Reduced values in underweight subjects normalize after An [18F]setoperone PET imaging study. Am J Psychiatry weight gain. Biol Psychiatry 23:102–105. 156:1029–1034. Kennedy SH, Javanmard M, Vaccarino FJ (1997): A review of Nieuwenhuys R, Voogd J, van Huijzen C (1988): The Human functional neuroimaging in mood disorders: Positron emis- Central Nervous System. Heidelberg, Germany: Springer- sion tomography and depression. Can J Psychiatry 42:467– Verlag. 475. O’Dwyer AM, Lucey JV, Russell GF (1996): Serotonin activity Klump KL, Bulik CM, Pollice C, Halmi KA, Fichter MM, in anorexia nervosa after long-term weight restoration: Re- Berrettini WH, et al (2000): Temperament and character in sponse to D-fenfluramine challenge. Psychol Med 26:353– women with anorexia nervosa. J Nerv Ment Dis 188:559– 359. 567. Palmer HD, Jones MS (1939): Anorexia nervosa as a manifes- Lambe EK, Katzman DK, Mikulis DJ, Kennedy SH, Zipursky tation of compulsive neurosis. Arch Neurol Psychiatry 856– RB (1997): Cerebral gray matter volume deficits after weight 860. recovery from anorexia nervosa. Arch Gen Psychiatry Pazos A, Probst A, Palacios JM (1987): Serotonin receptors in 54:537–542. the human brain—IV: Autoradiographic mapping of seroto- Leibowitz SF, Shor-Posner G (1986): Brain serotonin and eating nin-2 receptors. Neuroscience 21:123–139. behavior. Appetite 7(Suppl):1–14. Price JC, Lopresti BJ, Meltzer CC, Smith GS, Mason NS, Huang Lemaire C, Cantineau R, Guillaume M, Plenevaux A, Christiaens Y (2001): Analyses of [(18)F]altanserin bolus injection PET L (1991): Fluorine-18-altanserin: A radioligand for the study data. II: Consideration of radiolabeled metabolites in humans. of serotonin receptors with PET: Radiolabeling and in vivo Synapse 41:11–21. biologic behavior in rats. J Nucl Med 32:2266–2272. Reiman EM, Raichle ME, Butler FK, Herscovitch P, Robins E Lilenfeld LR, Kaye WH, Greeno CG, Merikangas KR, Plotnicov (1984): A focal brain abnormality in panic disorder, a severe K, Pollice C, et al (1998): A controlled family study of form of anxiety. Nature 310:683–685. anorexia nervosa and bulimia nervosa: Psychiatric disorders in first-degree relatives and effects of proband comorbidity. Rioux A, Fabre V, Lesch KP, Moessner R, Murphy DL, Arch Gen Psychiatry 55:603–610. Lanfumey L (1999): Adaptive changes of serotonin 5-HT2A receptors in mice lacking the serotonin transporter. Neurosci Logan J, Fowler JS, Volkow ND, Wolf AP, Dewey SL, Schlyer Lett 262:113–116. DJ, et al (1990): Graphical analysis of reversible radioligand binding from time-activity measurements applied to [N-11C- Roser W, Bubl R, Buergin D, Seelig J, Radue EW, Rost B methyl]-(-)-cocaine PET studies in human subjects. J Cereb (1999): Metabolic changes in the brain of patients with Blood Flow Metab 10:740–747. anorexia and bulimia nervosa as detected by proton magnetic resonance spectroscopy. Int J Eating Disord 26:119–136. Lopresti B, Holt D, Mason S, Huang Y, Ruszkiewicz J, Per- evuznik J, et al (1998): Characterization of the radiolabeled Roth BL, Berry SA, Kroeze WK, Willins DL, Kristiansen K metabolites of [18F]altanserin: Implications for kinetic mod- (1998): Serotonin 5-HT2A receptors: Molecular biology and eling. In: Carson R, Daube-Witherspoon M, Herscovitch P, mechanisms of regulation. Crit Rev Neurobiol 12:319–338. editors. Quantitative Brain Imaging with Positron Emission Sandor S, Leahy R (1997): Surface-based labeling of cortical Tomography. San Diego, CA: Academic Press, pp 293–298. anatomy using a deformable atlas. IEEE Trans Med Imaging Lucki I (1998): The spectrum of behaviors influenced by 16:41–54. serotonin. Biol Psychiatry 44:151–162. Saucier C, Morris SJ, Albert PR (1998): Endogenous seroto- Mann JJ (1999): Role of the serotonergic system in the patho- nin-2A and -2C receptors in Balb/c-3T3 cells revealed in genesis of major depression and suicidal behavior. Neuropsy- serotonin-free medium: Desensitization and down-regulation chopharmacology 21(2 Suppl):99S–105S. by serotonin. Biochem Pharmacol 10:1347–1357. 906 BIOL PSYCHIATRY G.K. Frank et al 2002;52:896–906

Schneider F, Weiss U, Kessler C, Muller-Gartner HW, Posse S, Sullivan PF (1995): Mortality in anorexia nervosa. Am J Psychi- Salloum JB, et al (1999): Subcortical correlates of differential atry 152:1073–1074. classical conditioning of aversive emotional reactions in Treasure J, Campbell I (1994): The case for biology in the social phobia. Biol Psychiatry 45:863–871. aetiology of anorexia nervosa. Psychol Med 24:3–8. Simansky KJ (1996): Serotonergic control of the organization of Tucker DM, Luu P, Pribram KH (1995): Social and emotional feeding and satiety. Behav Brain Res 73:37–42. self-regulation. Ann N Y Acad Sci 769:213–239. Smith GS, Price JC, Lopresti BJ, Huang Y, Simpson N, Holt D, Van der Linden G, van Heerden B, Warwick J, Wessels C, van et al (1998): Test-retest variability of serotonin 5-HT2A Kradenburg J, Zungu-Dirwayi N, Stein DJ (2000): Functional receptor binding measured with positron emission tomogra- brain imaging and pharmacotherapy in social phobia: Single phy and [18F]altanserin in the human brain. Synapse 30:380– photon emission computed tomography before and after 392. treatment with the selective serotonin reuptake inhibitor Spielberger CD, Gorsuch RL, Lushene RE (1970): STAI Manual citalopram. Prog Neuropsychopharmacol Biol Psychiatry for the State Trait Anxiety Inventory. Palo Alto, CA: Consult- 24:419–438. ing Psychologists Press. Walsh BT, Devlin MJ (1998): Eating disorders: Progress and Spinks TJ, Jones T, Bailey DL, Townsend DW, Grootoonk S, problems. Science 280:1387–1390. Bloomfield PM (1992): Physical performance of a positron tomograph for brain imaging with retractable septa. Phys Med Ward A, Brown N, Lightman S, Campbell IC, Treasure J (1998): Biol 37:1637–1655. Neuroendocrine, appetitive and behavioural responses to d-fenfluramine in women recovered from anorexia nervosa. Srinivasagam NM, Kaye WH, Plotnicov KH, Greeno C, Weltzin Br J Psychiatry 172:351–358. TE, Rao R (1995): Persistent perfectionism, symmetry, and exactness after long-term recovery from anorexia nervosa. Wolfe BE, Metzger ED, Jimerson DC (1997): Research update Am J Psychiatry 152:1630–1634. on serotonin function in bulimia nervosa and anorexia nervosa. Psychopharmacol Bull 33:345–354. Staley JK, Malison RT, Innis RB (1998): Imaging the Seroto- nergic System: Interactions of neuroanatomical and func- Woods RP, Cherry SR, Mazziotta JC (1992): Rapid automated tional abnormalities of depression. Biol Psychiatry 44:354– algorithm for aligning and reslicing PET images. J Comput 549. Assist Tomogr 16:620–633. Staley JK, Van Dyck CH, Tan PZ, Al Tikriti M, Ramsby Q, Woods RP, Mazziotta JC, Cherry SR (1993): MRI-PET regis- Klump H, et al (2001): Comparison of [(18)F]altanserin and tration with automated algorithm. J Comput Assist Tomogr [(18)F]deuteroaltanserin for PET imaging of serotonin(2A) 17:536–546. receptors in baboon brain: Pharmacological studies. Nucl Med Yatham LN, Liddle PF, Shiah IS, Scarrow G, Lam RW, Adam

Biol 28:271–279. MJ, et al (2000): Brain serotonin2 receptors in major depres- Stockmeier CA (1997): Neurobiology of serotonin in depression sion: A positron emission tomography study. Arch Gen and suicide. Ann N Y Acad Sci 836:220–232. Psychiatry 57:850–858. Strober M, Freeman R, Lampert C, Diamond J, Kaye W (2000): Zald DH, Kim SW (1996): Anatomy and function of the orbital Controlled family study of anorexia nervosa and bulimia frontal cortex, II: Function and relevance to obsessive- nervosa: Evidence of shared liability and transmission of compulsive disorder. J Neuropsychiatry Clin Neurosci partial syndromes. Am J Psychiatry 157:393–401. 8:249–261. Strupp BJ, Weingartner H, Kaye W, Gwirtsman H (1986): Zalla T, Koechlin E, Pietrini P, Basso G, Aquino P, Sirigu A, Cognitive processing in anorexia nervosa: A disturbance in Grafman J (2000): Differential amygdala responses to win- automatic information processing. Neuropsychobiology ning and losing: A functional magnetic resonance imaging 15:89–94. study in humans. Eur J Neurosci 12:1764–1770.