Compulsive Disorder Have Increased 5-HT2A Receptor Binding in the Caudate Nuclei

Home , Altanserin, SCAN, Tic

ARTICLE International Journal of Neuropsychopharmacology (2005), 8, 391–401. Copyright f 2005 CINP doi :10.1017/S1461145705005055 Patients with obsessive–compulsive disorder have increased 5-HT2A receptor binding in the caudate nuclei

Karen H. Adams1, Elsebeth S. Hansen2, Lars H. Pinborg1, Steen G. Hasselbalch1, Claus Svarer1, Søren Holm3, Tom G. Bolwig2 and Gitte M. Knudsen1 1 Neurobiology Research Unit (NRU), University Hospital of Copenhagen, Denmark 2 The Department of Psychiatry, University Hospital of Copenhagen, Denmark 3 PET and Cyclotron Unit, University Hospital of Copenhagen, Denmark Downloaded from https://academic.oup.com/ijnp/article/8/3/391/910011 by guest on 24 September 2021

Abstract

The pharmacological efficacy of serotonergic-acting drugs suggest that patients with obsessive– compulsive disorder (OCD) may have alterations in their cerebral serotonergic (5-HT) receptor system, and previous neuroimaging studies of OCD patients have shown abnormalities in several fronto- subcortical regions. In this study we investigated cerebral 5-HT2A receptor binding in 15 untreated OCD patients and in 15 age- and gender-matched healthy volunteers by magnetic resonance imaging and [18F]altanserin positron emission tomography (PET). Eleven of the patients were rescanned with PET after receiving treatment with a selective serotonin reuptake inhibitor (SSRI). The distribution volumes of specific tracer binding (DV3k) were calculated for 12 brain regions, and comparisons were made between: (1) healthy volunteers vs. untreated OCD patients, (2) healthy volunteers vs. treated OCD patients, and (3) OCD patients before and during treatment. When comparing the distribution volume for specific fronto- subcortical brain regions, significantly higher values were recorded in the caudate nuclei in OCD patients

(DV3k : 0.24¡0.14) compared to the healthy control group (DV3k: 0.15¡0.13) (p<0.05, Wilcoxon matched- pairs test). This diﬀerence between groups was not present after treatment with SSRIs. There was no correlation between the severity of OCD symptoms and 5-HT2A receptor binding. An increase in 5-HT2A receptor binding is found in the caudate nuclei of untreated patients with OCD. The up-regulation in

5-HT2A receptors might be compensatory for a lack of serotonin in the feedback loop between the thalamus and orbito-frontal cortex, the caudate nuclei, and the globus pallidus. Received 17 March 2004; Reviewed 20 June 2004; Revised 15 September 2004; Accepted 5 October 2004

Key words: OCD, positron emission tomography (PET), serotonin 5-HT2A receptors.

Introduction equally and it has a mean age of onset of approx. 20 yr of age. Obsessive–compulsive disorder (OCD) is a psychiatric OCD has been classified as an anxiety disorder by disorder, which affects 2–3% of the population both the American Psychiatric Association (DSM-IV; worldwide (Karno et al., 1988; Weissman et al., 1994). APA, 1994), and the World Health Organization (ICD- It has been estimated that up to 10% of patients 10; WHO, 1992–1994). OCD symptoms include recur- from outpatient psychiatric clinics suffer from rent, unwanted thoughts (obsessions) and conscious, OCD, making it one of the most common diagnoses in ritualized acts (compulsions) usually attempting to psychiatry, after phobias, substance abuse and, major deal with the apprehension generated by the obses- depression. The disease attacks men and women sions. OCD is frequently a chronic and debilitating condition as the obsessions and compulsions engage the patient for numerous hours every day. Address for correspondence : Professor G. M. Knudsen, Neurobiology To maximize treatment outcome in OCD, various Research Unit N9201, University Hospital of Copenhagen, 9 treatment approaches need to be combined. There is Blegdamsvej, Copenhagen, DK-2100 Denmark. Tel.: (+45) 3545 6720 Fax :(+45) 3545 6713 extensive evidence that indicates that selective sero- E-mail : [email protected] tonin re-uptake inhibitors (SSRIs) and behavioural 392 K. H. Adams et al. therapy applying the principles of exposure and re- from post-mortem brain studies or imaging studies. sponse prevention (Cottraux et al., 2001; Lindsay et al., In this study we investigated whether OCD patients

1997), are highly eﬀective in reducing the symptoms of show alterations in their 5-HT2A receptor system OCD. Multiple randomized, double-blind, placebo- compared to normal controls, and whether such ab- controlled studies have demonstrated the eﬃcacy of normalities could be reversed by successful medical SSRIs alone (Jenike et al., 1990a,b; Ravindran et al., treatment. We hypothesized that OCD patients have

1999; Zohar and Judge, 1996). alterations in their density of serotonin 5-HT2A recep- The cause of OCD remains unknown. However, tors in regions of the fronto-subcortical brain circuit, family-genetic studies suggest that there may be mul- namely the gyrus rectus (part of the orbito-frontal tiple aetiological subtypes (Pauls et al., 1986; Pauls cortex), anterior cingulate cortex, thalamus, and the and Leckman, 1986). A review of the OCD functional caudate nuclei. It was also hypothesized that these brain-imaging literature reveals a remarkable amount changes would normalize following successful SSRI

of data suggesting abnormal functioning along treatment, since in PET studies of major depression a Downloaded from https://academic.oup.com/ijnp/article/8/3/391/910011 by guest on 24 September 2021 specific, fronto-subcortical brain circuits comprising reduction in 5-HT2A binding is found following the orbito-frontal cortex, cingulate cortex, caudate medical treatment (Attar-Levy et al., 1999; Meyer et al., nuclei, and the thalamus. Previous positron emission 2001). tomography (PET) studies have examined cerebral glucose consumption in OCD patients in comparison Method to healthy controls and in OCD patients before and Patients and controls after treatment. In untreated OCD, there is evidence of an elevated glucose metabolism in the fronto- Fifteen outpatients, eight women and seven men, with subcortical brain circuit (Baxter et al., 1987, 1988; a mean age of 38 yr (range 18–73 yr) were recruited Perani et al., 1995; Saxena et al., 2001; Swedo et al., from the Anxiety Disorder Unit in the University 1989), and a normalization of glucose consumption in Department of Psychiatry, Copenhagen. Diagnosis response to pharmacological treatment (Baxter et al., and clinical evaluation were made on the basis of a 1992; Hansen et al., 2002; Perani et al., 1995; Saxena semi-structured interview designed for the study by et al., 1999, 2002; Swedo et al., 1992). Further, during two trained psychiatrists (E.S.H. and T.G.B.). All provoked OCD symptoms, brain activation with patients fulfilled the DSM-IV and the ICD-10 criteria elevated cerebral blood flow in the fronto-subcortical for OCD, and the severity of OCD symptoms were brain circuit has been found (McGuire et al., 1994; assessed by the Yale–Brown Obsessive–Compulsive Rauch et al., 1994). Scale (YBOCS; Goodman et al., 1989). For assessment In addition to the abnormal functioning along of comorbidity with other psychiatric conditions at specific fronto-subcortical brain circuits in OCD, it has the time of study, patients were interviewed according been suggested that OCD is caused by a dysfunction to the Brief Psychiatric Rating Scale (BPRS) and in the serotonergic system, since SSRIs have a clinical Hamilton Depression Scale (HAMD). During the effect in OCD. Furthermore, recent data suggest that interviews we diagnosed the following Axis II dis- the activation of 5-HT2A and/or 5-HT2C receptors may orders: Two had dependent personality disorder, two be important for the improvement of OCD symptoms. had obsessive–compulsive personality disorder and

Thus, some hallucinogens with 5-HT2 agonist proper- one patient had histrionic personality disorder. ties have been reported to have a beneﬁcial eﬀect, e.g. A general somatic and neurological examination LSD and psilocybin (Delgado and Moreno, 1998; was performed in all subjects. There was no suspicion Hanes, 1996; Moreno and Delgado, 1997). In addition, of substance abuse in any of the subjects and labora- administration of the serotonergic agent m-chloro- tory screening for drug or alcohol were thus not per- phenylpiperazine (m-CPP), which binds to 5-HT1/ formed. Subjects were rated with YBOCS and HAMD

5-HT2 receptors, have been shown to exacerbate symp- at the time of the PET scan both before and after toms in some OCD patients (Hollander et al., 1992). treatment. Responders to treatment were deﬁned as This neuroendocrine and behavioural eﬀect of m-CPP those that rated either ‘much improved’ or ‘very in patients with OCD are, however, antagonized by much improved’ on item 18 of the YBOCS at the time the non-selective 5-HT antagonist metergoline, sug- of the second PET scan. gesting that the response of m-CPP is mediated by Four patients did not complete the post-treatment

5-HT1 and/or 5-HT2 receptors (Pigott et al., 1993). PET-study; one patient was excluded from the study So far, there has been a surprising lack of data on because of a concurrent depression, whilst the other the serotonergic receptor system with no contribution three did not wish to undergo a second scan. 5-HT2A receptor binding in patients with OCD 393

Table 1. Patient material, demographic characteristics and clinical data

Prior treatment Patient Duration YBOCS Study Age of illness Predominant Medi- Psycho- Family no. Sex (yr) (yr) symptoms Before After cation therapy Comorbidity history

21 F 49 30 Checking 31 15*** ++ Depression OCD 28 F 54 39 Ritualistic eating 31 x+ + Anorexia x 30 F 21 10 Mixed 28 10*** xx x x 31 F 18 10 Ritualistic eating 30 2*** x+ x x 32 F 24 1 Obsessions 14 4*** ++ Anxiety x 40 F 56 21 Washing 24 15*** ++ Depression x

51 F 40 15 Washing 25 xx + single phobia OCD Downloaded from https://academic.oup.com/ijnp/article/8/3/391/910011 by guest on 24 September 2021 53 F 22 5 Checking 32 10*** x+ x Depression 34 M 43 20 Washing 40 xx x x OCD 37 M 73 25 Checking 36 22*** x+ Depression Depression 39 M 44 15 Checking 38 x+ + Depression OCD 43 M 43 12 Washing 30 8*** xx x x 46 M 25 16 Obsessions 24 16*** xx x OCD 47 M 29 14 Checking 20 6*** +x Depression x 59 M 22 11 Washing 30 11*** ++ x OCD

YBOCS score – severity of OCD: 0–7 subclinical; 8–15 mild; 16–23 moderate; 42–31 severe; 32–40 extreme. Median YBOCS score before treatment with SSRI=30 (range 14–40). Median YBOCS score during treatment with SSRI=10 (range 2–22). *** Indicates that the decrease in YBOCS scores is highly signiﬁcant (t=5.402, f=14, p<0.01).

All patients were declared physically healthy and OCD and two patients had a family history of all but one patient, who had suffered from migraine, depression among first-degree relatives. had no previous history of neurological illness, In 11 patients, previous attempts to treat OCD including Tourette’s Syndrome or other tic disorders, symptoms had been made, either by contact with a nor had they experienced any major head trauma. psychiatrist, a general practitioner or a psychologist. The duration of OCD symptomatology ranged from The 10 patients who had received psychotherapy had 1 to 39 yr with a mean of 16 yr (S.D.=9.7 yr). The age of stopped several months before the onset of the study, onset in one patient was 35 yr and in another 48 yr. and the seven patients who had been pharmacologi- All patients had obsessions as well as compulsions cally treated had been completely drug free at least of varying severity. Five patients had checking be- 4 wk prior to entering the study. Common for all haviour as a their predominant symptom, five wash- patients, however, was that previous attempts to treat ing rituals, one patient a mixed symptomatology with within the last 12 months prior to the study had been repetition rituals, counting and need for symmetry unsuccessful. Clinical data are presented in Table 1. and exactness, and two patients suffered from ritual- Control subjects were recruited through newspaper istic eating. None of the two latter patients suffered advertisements. Fifteen gender- and age-matched from anorexia nervosa; their OCD symptoms were healthy volunteers, eight women and seven men, with related to way of chewing, arranging elements of food, a mean age of 39 yr (range 21–75 yr) were included in dishes, etc. in a ritualistic order. the study. Before inclusion all controls were inter- Eight of the patients had a history of comorbidity: viewed by a psychiatrist using the same scales as those Five patients had suffered from depression, one pa- used for the patients. One control subject was ex- tient had had a single phobia, one had suffered from cluded due to a YBOCS score of 4. None of the subjects anorexia nervosa and one patient had had generalized had past or present psychiatric or neurological dis- anxiety symptoms. However, none of these comorbid orders. Furthermore, none of the volunteers were conditions had been manifest for at least 6 months receiving medical treatment acting on the central ner- prior to the study. Six patients had a family history of vous system. 394 K. H. Adams et al.

Written informed consent was obtained from bodyweight [18F]altanserin, given as a bolus-infusion patients and controls according to the Declaration of ratio of 1.75 h, according to Pinborg et al. (2003). Helsinki II, and the study was approved by the Ethics PET scans were performed with an 18-ring GE Committee of Copenhagen and Frederiksberg [(KF) Advance scanner (GE, Milwaukee, WI, USA) operat- 02–058/99]. ing in 3D acquisition mode, producing 35 image slices with an interslice distance of 4.25 mm. The total axial Treatment field of view was 15.2 cm with an approximate in- plane resolution of 5 mm. The technical specifications The patients received SSRI treatment daily, either have been described by DeGrado et al. (1994) and paroxetine (60–80 mg), sertraline (50–150 mg), fluox- Lewellen et al. (1996). etine (60–80 mg), or citalopram (60–80 mg). Plasma Ninety minutes after the bolus injection of drug levels were not monitored, but all patients [18F]altanserin the subjects were placed in the scanner. showed good compliance and relatively few had side-

Subjects were aligned in the scanner using a laser Downloaded from https://academic.oup.com/ijnp/article/8/3/391/910011 by guest on 24 September 2021 effects. One patient refused pharmacological treatment system so that the detectors were parallel to the orbito- and received counselling only. The patients were not meatal line, and positioned to include the cerebellum allowed to and did not receive psychotherapy from in the field of view using a short 2-min transmission outside the study. At inclusion, the patients received scan. An individual head holder was made to ensure weekly consultations and later, after initializing drug relative immobility. All subjects were scanned in a treatment, they had consultations every 4 wk. The re- resting state. scans were performed at the earliest 12 wk after start- Dynamic 3D emission scans (five frames of 8 min) ing the pharmacological therapy (median 26.5 wk, started 120 min after administration of [18F]altanserin. range 12–38 wk). Prior to this scan a 10-min transmission scan was obtained for correction of tissue attenuation, using MRI studies retractable 68Ge/68Ga pin sources. The transmission All subjects had a structural MRI scanning performed scans were corrected for tracer activity by a 5-min with a 1.5 T Vision scanner (Siemens, Erlangen, emission scan performed in 2D mode. Germany) using a 3D MPRAGE sequence (TI/TE/ Data were reconstructed into a sequence of 128 128 35 voxel matrices, each voxel measuring TR=100/4.4/11.4 ms, flip angle 8x). The images were r r acquired as sagittal plane scans with a spatial resol- 2.0r2.0r4.25 mm, with software provided by the 3 manufacturer (GE, Milwaukee, WI, USA). A 3D ution of 1.1r1.2r1.2 mm . The number of planes was re-projection algorithm with a transaxial Hann filter 158 and the in-plane matrix dimensions were 256r192. (6 mm) and an axial ramp filter (8.5 mm) was applied. Corrections for dead-time, attenuation, and scatter [18F]altanserin PET studies were performed.

PET imaging of the regional distribution of 5-HT2A Five venous blood samples were drawn at mid-scan receptors was done with [18F]altanserin, which is a times 4, 12, 20, 28 and 36 min after starting the dy- 18 5-HT2A antagonist. [ F]altanserin has high affinity for namic scanning sequence. The samples were immedi- the 5-HT2A receptor with a Ki=0.13 nM, which is at ately centrifuged, and 0.5 ml of plasma was counted in least a 20-fold greater than Ki for other 5-HT subtypes a well counter for determination of radioactivity. Three (Tan et al., 1999). The radiosynthesis of [18F]altanserin of the five blood samples drawn at 4, 20 and 36 min was prepared according to a previously described were also analysed for percentage of parent compound method by Lemaire et al. (1991). Quality control was ([18F]altanserin) using reverse-phase HPLC following performed using analytical TLC and HPLC. For each the procedure described by Pinborg et al. (2003). PET study 0.4–3.5 GBq of [18F]altanserin was produced with a radiochemical purity greater than 98% and a Data analysis mean specific activity of 169 GBq/mmol (range 71–300 GBq/mmol). Reconstructed and decay-corrected PET images and Catheters were inserted in both cubital veins for MRI images were transferred to Hewlett Packard tracer injection and blood sampling respectively. UX workstations. Images were co-registered using [18F]altanserin was administrated as a combination of a Matlab-based program, IPS (Interactive Point a bolus injection and a continuous infusion to obtain Selection), developed in-house, based on selection of tracer equilibrium both in plasma and brain tissue. at least six common anatomical locations in the two Subjects received a maximum dose of 3.7 MBq/kg images. The transformation between the two sets of 5-HT2A receptor binding in patients with OCD 395

z = 25.5 mm z = 29.8 mm z = 34.0 mm z = 38.2 mm z = 42.5 mm

z = 46.8 mm z = 51.0 mm z = 55.2 mm z = 59.5 mm z = 63.8 mm

z = 68.0 mm z = 72.2 mm z = 76.5 mm z = 80.8 mm z = 85.0 mm Downloaded from https://academic.oup.com/ijnp/article/8/3/391/910011 by guest on 24 September 2021

Figure 1. Co-registered [18F]altanserin PET image and MRI image including all Regions of Interest. Identiﬁcation numbers are shown the ﬁrst time a region appears. (1) Orbito-frontal cortex, (2) cerebellum, (3) hippocampus, (4) ventral lateral frontal cortex, (5) insula, (6) lentiform nuclei, (7) anterior cingulate, (8) caudate nuclei, (9) temporal cortex, (10) dorsal lateral prefrontal cortex, (11) thalamus, (12) parietal cortex. points was estimated in a least-squares sense, and the volume (DV) can be calculated as the ratio of tissue MRI image was re-sliced into PET space. radiotracer concentration to plasma radiotracer con-

Regions-of-interest (ROI) analysis was performed centration. In receptor studies the DV of a ROI (DVROI) using a Matlab-based program, Editroi, also devel- represents the sum of the specifically bound DV (DV3) oped in house. ROIs were drawn on the co-registered and the non-specifically bound (DV2). DVROI is related MRI images and transferred to the PET images. to binding parameters as follows: Twelve regions were selected: cerebellum, gyrus rec- 0 tus (part of the orbito-frontal cortex), ventral lateral Bmax DVROI=DV2+DV3=DV2+f1 (ml=ml), (1) frontal cortex, hippocampus, anterior cingulate, Kd insula, caudate nuclei, lentiform nuclei, thalamus, where f1 is the free tracer fraction (not bound to plasma temporal cortex, parietal cortex and dorsal lateral proteins) assuming that the concentration of free ligand prefrontal cortex (see Figure 1). The ROIs were deter- in tissue water equals the concentration in plasma mined by inspection of the MRI images with reference water, Bkmax (nM) is the concentration of receptors to the brain atlas of Talairach and Tournoux (1988), available for binding, and Kd (nM) is the affinity con- and two investigators (K.H.A. and S.G.H.) blinded stant of [18F]altanserin. From eqn (1) it can be seen that to the clinical diagnosis, agreed on the ROI positions. DVROI is a linear function of Bkmax provided that Kd is A mean count density was obtained for each anatom- the same. Due to overall physicochemical constancy of ical structure by averaging values across all slices in the brain tissue, DV2 is often assumed to equal the DV which the structure appeared. Count densities were of a ROI devoid of receptors (DVref). Inserting DVref= expressed in MBq/cc. DV2 and rearranging eqn (1) yields the outcome Decay-corrected plasma curves were corrected measure used in this study: for the presence of radiolabelled metabolites of 0 [18F]altanserin, as determined by the HPLC analysis. 0 Bmax DV3=f1 =DVROIxDVref (ml=ml), (2) Brain and plasma curves were visually inspected to Kd ensure the presence of steady-state in both brain and which also can be given as eqn (3) using cerebellum as plasma activity. a reference: The steady-state method requires that equilibrium is present both in blood and brain tissue (Lassen, 0 CROIxCcerebellum DV3= (ml=ml): (3) 1992). When steady-state is present, the distribution Cplasma 396 K. H. Adams et al.

¡ Statistical analysis Table 2. Distribution volume (DV3k S.D.) for all regions of interest in healthy controls and patients with OCD (pre-treatment) Statistical analysis of the regional DV3k data for gyrus rectus, anterior cingulate cortex, thalamus and the caudate nuclei were performed using Wilcoxon OCD patients matched-pairs tests, with a signiﬁcance level at p<0.05. Controls pre-treatment Three separate comparisons were analysed: (1) healthy Region (n=15) (n=15) volunteers vs. untreated OCD patients (n=15), (2) Gyrus rectus 1.37¡0.42 1.36¡0.51 healthy volunteers vs. treated patients (n=11) and (3) Anterior cingulate 1.29¡0.47 1.31¡0.51 OCD patients before and during treatment (n=11). A Caudate nuclei* 0.15¡0.14 0.23¡0.15 paired test was used, as the controls and patients were Thalamus 0.29¡0.14 0.33¡0.15 carefully matched to counteract the eﬀect of ageing Temporal cortex 1.41¡0.49 1.44¡0.50

(Meltzer et al., 1998; Rosier et al., 1996; Sheline et al., Downloaded from https://academic.oup.com/ijnp/article/8/3/391/910011 by guest on 24 September 2021 ¡ ¡ 2002) and gender (Biver et al., 1996) and for the evalu- Parietal cortex 1.39 0.50 1.40 0.51 Ventral lateral frontal cortex 1.15¡0.43 1.19¡0.49 ation of treatment, patients acted as their own controls. Dorsal lateral prefrontal cortex 1.26¡0.48 1.29 ¡.0.49 Similarly, Wilcoxon matched-pairs tests, corrected for Insula 1.31¡0.47 1.32¡0.48 multiple comparisons with a Bonferoni correction Hippocampus 0.45¡0.27 0.50¡0.23 (p<0.007) were performed for the seven ROIs not Lentiform nuclei 0.47¡0.17 0.48¡0.18 included in the fronto-subcortical circuit when looking Cerebellum 1.60¡0.46 1.52¡0.45 for global changes in the brain.

The difference between DV3k for all ROIs in patients/ The ROIs above the line belong to the fronto-subcortical brain controls or patients before and after treatment were circuit. analysed by linear mixed models. It was first tested * Indicates that DV3k for the caudate nuclei are significantly whether the mean difference was the same in all ROIs higher for OCD patients (pre-treatment) than for the controls and then whether this common difference was zero. (p<0.05, Wilcoxon matched pairs test). The non-specific binding defined as: C /C is shown at the bottom The difference was allowed to be correlated within cerebellum plasma of the table. patients or patients/controls respectively. Correlation analysis (ANCOVA) with age as a co- variate was conducted to evaluate possible treatment Results effects. Correlations were evaluated between the Clinical data YBOCS scores and per cent change in DV3kOCD for gyrus rectus, anterior cingulate, caudate nuclei and The 15 patients participating in the study had the fol- thalamus, as determined before and after treatment. lowing clinical scores (median¡S.D.): BPRS 1.0¡1.9 In order to identify clusters with particularly in- and HAMD 6.0¡2.9, which correspond to values creased/decreased binding, e.g. responders vs. non- found in the healthy subjects. The YBOCS score in the responders, Mann–Whitney tests were performed unmedicated patients were 30¡6.8. Following treat- with a significance level of p<0.05 and p<0.007, as ment, the YBOCS score dropped by 40% (t=5.402, explained above. f=14, p=<0.001) (see Table 1). Four patients were As an additional explorative analysis to identify non-responders with a post-treatment YBOCS score possible ROIs involved, a statistical parametric map- >24, while the remainder responded to treatment ping (SPM 99; Wellcome Department of Cognitive with YBOCS scores <22. All patients that completed

Neurology, London, UK) analysis was performed. DV3k the second PET scan (n=11) were responders. None of images were generated [eqn (3)] and spatially nor- the controls scored on any of the psychiatric scales. malized into Montreal Neurological Institute stereo- 5-HT receptor binding tactic space using the corresponding MRI image and 2A

SPM 99 software. Normalized DV3k images were Tables 2–4 show the distribution volumes (DV3k) for smoothed with a gaussian filter to 12 mm full-width [18F]altanserin binding in all ROIs. Untreated OCD half maximum (SPM 99). Hypothesis testing was patients have a higher binding of [18F]altanserin in performed on a voxelrvoxel basis, comparing the both the left and the right caudate nucleus when same three contrasts as mentioned above using an compared to a matched control group ( f=14, p<0.05) ANCOVA test with age as a nuisance in SPM 99, at (see Figure 2). No significant differences were found p<0.05 (corrected). Also paired t tests at p<0.05 (cor- for DV3k in any of the other regions for the conditions rected) were performed in SPM 99. tested. 5-HT2A receptor binding in patients with OCD 397

Table 3. Distribution volume (DV3k¡S.D.) for all regions of Table 4. Distribution volume (DV3k¡S.D.) for all regions of interest in healthy controls and patients with OCD (post- interest in the OCD patients before and after treatment with treatment) SSRI

OCD patients OCD patients OCD patients Controls post-treatment pre-treatment post-treatment Region (n=11) (n=11) Region (n=11) (n=11)

Gyrus rectus 1.42¡0.39 1.29¡0.50 Gyrus rectus 1.42¡0.56 1.29¡0.50 Anterior cingulate 1.33¡0.45 1.33¡0.53 Anterior cingulate 1.36¡0.58 1.33¡0.53 Caudate nuclei 0.18¡0.15 0.28¡0.20 Caudate nuclei 0.29¡0.16 0.28¡0.20 Thalamus 0.31¡0.14 0.33¡0.13 Thalamus 0.36¡0.16 0.33¡0.13

Temporal cortex 1.46¡0.46 1.44¡0.54 Temporal cortex 1.47¡0.58 1.44¡0.54 Parietal cortex 1.46¡0.51 1.41¡0.54 Parietal cortex 1.45¡0.59 1.41¡0.54 Downloaded from https://academic.oup.com/ijnp/article/8/3/391/910011 by guest on 24 September 2021 Ventral lateral frontal 1.23¡0.42 1.18¡0.51 Ventral lateral 1.24¡0.53 1.18¡0.51 cortex frontal cortex Dorsal lateral prefrontal 1.34¡0.45 1.36¡0.54 Dorsal lateral 1.35¡0.55 1.36¡0.54 cortex prefrontal cortex Insula 1.38¡0.46 1.26¡0.52 Insula 1.35¡0.55 1.26¡0.52 Hippocampus 0.49¡0.29 0.51¡0.26 Hippocampus 0.54¡0.25 0.51¡0.26 Lentiform nuclei 0.48¡0.15 0.48¡0.18 Lentiform nuclei 0.50¡0.20 0.48¡0.18 Cerebellum 1.53¡0.31 1.38¡0.43 Cerebellum 1.41¡0.48 1.38¡0.43

(Ccerebellum/Cplasma) The ROIs above the line belong to the fronto-subcortical brain circuit. The non-speciﬁc binding deﬁned as: Ccerebellum/ The ROIs above the line belong to the fronto-subcortical brain

Cplasma is shown at the bottom of the table. circuit. The non-speciﬁc binding deﬁned as: Ccerebellum/

Cplasma is shown at the bottom of the table. Statistical testing using linear mixed models of differences in DV3k in controls and OCD patients showed SSRI treatment, and no correlation between duration that all regions could be assumed to have the same of SSRI treatment and differences in 5-HT2A binding of expected difference between patients and controls. OCD patients before and after treatment. Exclusion of Furthermore, this shared expected difference was the patient who received clinical counselling only found not to be significantly different from zero. The (n=10) did not alter the results. significance of these fixed effects was tested in a model There was no statistically significant correlation that took the large correlation between differences in between YBOCS score and DV3k for the gyrus rectus,

DV3k within the same matched pair into account. anterior cingulate, caudate nuclei, and thalamus; nor The distribution volumes are generally higher (on were there any statistically signiﬁcant correlations average 9%) in all ROIs, in unmedicated OCD patients between the relative (before and after treatment) compared to the matched control group (n=15) and changes in these variables (data not shown). slightly higher (4%) for the treated OCD patients The results obtained using the SPM analysis of DV3k compared to the matched controls (n=11). For the images were similar to the results obtained by the ROI

OCD patients an averaged decline in all ROIs of DV3k of analysis, in that the SPM analysis did not reveal any 3% occurred after treatment (n=11). None of these statistically different clusters/ROI between any of the changes were, however, statistically significant. No groups tested. Due to the restrictive statistics involved differences were found in the non-specific binding, as in testing without any a-priori hypothesis, SPM defined from the cerebellar DV3k (Wilcoxon matched- analysis failed to identify that untreated OCD patients pairs test). have higher [18F]altanserin binding in the caudate

There was no difference in 5-HT2A binding nuclei compared to controls. between patients with a family history of depression and/or OCD, and patients without such a family his- Discussion tory. In addition, there was no difference between SSRI drug-naive patients and those who on previous occa- We find that in unmedicated patients with OCD, DV3k 18 sions had received SSRI therapy. Further, there was no for [ F]altanserin binding to 5-HT2A receptors in the difference between responders and non-responders to caudate nuclei is increased compared to an age- and 398 K. H. Adams et al.

0.5 p < 0.05 0.4 ' 3 0.3

DV 0.24 ± 0.14 0.2 0.15 ± 0.13 0.1

0.0 Controls OCD patients Downloaded from https://academic.oup.com/ijnp/article/8/3/391/910011 by guest on 24 September 2021

Figure 2. Distribution volume (DV3k) for the caudate nuclei in controls and untreated OCD patients (n=15). The mean DV3k value for the controls is 0.15¡0.13 (S.D.) and the mean DV3k value for the OCD patients is 0.24¡0.14 (S.D.). p<0.05 indicates that the

DV3k for the caudate nuclei is significantly higher than in OCD patients when compared to healthy volunteers. gender-matched control group. It should be pointed OCD patients have been found to have a significantly out that the relatively low density of 5-HT2A receptors higher glucose metabolism in this region compared to in the caudate nuclei complicates the detection of alter- a control group (Baxter et al., 1987, 1988). Further, ations present, because of the relative uncertainty in imaging studies in OCD have demonstrated a de- the determinations. Test–retest data for [18F]altanserin creased metabolic activity in the caudate nuclei, both after bolus injection is approx. 10% (Smith et al., 1998) in response to treatment with SSRI and behavioural which is in the same order of magnitude as for the therapy (Baxter et al., 1992; Hansen et al., 2002; Saxena bolus-infusion approach (Pinborg, Adams, Knudsen, et al., 1999, 2002; Schwartz et al., 1996), and also an unpublished data). It is unknown to what extent there increased cerebral blood flow was found in the is a biological day-to-day variation, and therefore, it is caudate nucleus in symptom provocation studies not possible to assess the variability that stems from (McGuire et al., 1994; Rauch et al., 1994). Our data methodological instability alone. This possibly ex- support and extend the previously reported pathology plains why the small and statistically insignificant in the caudate nuclei of OCD. decrease in DV3k of the caudate nuclei following SSRI Our finding of an increased distribution volume treatment caused a loss of statistically significant dif- may either be caused by an increase in the regional ference between groups. We cannot exclude that the density of 5-HT2A receptors or by a decrease in radio- latter finding is an effect of the limited sample size in ligand affinity. The latter situation could possibly be the within-subject study of SSRI treatment (n=11) and induced in the case of a decrease in the synaptic con- that the study may, therefore, lack power to detect centration of competing serotonin. However, since 18 effects of SSRI treatment on the number of 5-HT2A [ F]altanserin binding remains unaltered during in- receptors. The statistical difference between un- creased concentrations of synaptic serotonin (Pinborg medicated OCD patients and controls was observed et al., in press), we conclude that the difference is with the ROI analysis only, whereas both the ROI and caused by an up-regulation of available 5-HT2A bind- SPM analyses confirmed that there are no global alter- ing sites (Bmax) in the caudate nuclei. Our conclusion ations in the 5-HT2A receptor density in OCD patients as that our findings are due to 5-HT2A receptor up-regu- compared to gender- and age-matched controls. SPM, lation are supported by post-mortem studies that however, takes a data-driven approach which might demonstrated increased density of 5-HT2A receptors in be useful for the generation of hypotheses but is less the prefrontal cortex of suicide victims, and depressed suitable for detection of subtle differences in pre- subjects who died of natural causes (Arango et al., defined regions. Further, SPM involves filtering with a 1997; Arora and Meltzer, 1989; Hrdina et al., 1993),

12-mm gaussian filter and is, thus, less likely to pick suggesting that 5-HT2A receptors up-regulate in up smaller signal changes in regions located close to response to a defect in 5-HT neurotransmission. the ventricles, because of partial volume effects. A potential confounding factor, the choice of refer- The location of changes to the caudate nuclei is in ence region, needs to be considered with respect to agreement with findings from other groups where DV3k. It is required that the reference region represents 5-HT2A receptor binding in patients with OCD 399 non-specific binding only, or at least has negligible Acknowledgements specific binding. In a previous paper it was found that The authors thank Karin Stahr and the staff at the PET cerebellar 5-HT2 binding density was approximately A Centre at Rigshospitalet, Copenhagen, for their assist- one third of the density in the receptor-rich region ance. Thanks are also due to The Danish Health prefrontal cortex (Eastwood et al., 2001). In our lab- Research Council, the 1991 Pharmacy Foundation, the oratory, the suitability of the cerebellum as a reference Health Insurance Fund, the Lundbeck Foundation, region was evaluated in [18F]altanserin PET with Lægernes forsikringsforening af 1891 and Fonden af within-scan displacement with cold ketanserin 1870 for funding the project. In addition, thanks are (Pinborg et al., 2003), showing a complete blockade of due to the John and Birthe Meyer Foundation for the all ROIs binding except for the cerebellum, which donation of the Cyclotron and PET scanner. remained unchanged following displacement. Based on these data, the cerebellum seems to serve as a 18

suitable reference region in clinical [ F]altanserin PET Downloaded from https://academic.oup.com/ijnp/article/8/3/391/910011 by guest on 24 September 2021 Statement of Interest studies. In addition, there was no diﬀerence between the cerebellar [18F]altanserin binding of OCD patients None. and control subjects. A previous study has shown signiﬁcantly higher

5-HT2A binding potential in men than in women, References especially in the frontal and cingulate cortices (Biver APA (1994). Diagnostic and Statistical Manual of Mental et al., 1996). Furthermore, it has been reported that Disorders (4th edn). Washington, DC: American Psychiatric differences in hormonal changes in women may Press. influence 5-HT2A receptors (Moses et al., 2000). To as- Arango V, Underwood MD, Mann JJ (1997). Postmortem sess any possible interference we also conducted findings in suicide victims. Implications for in vivo imaging comparisons between males only (n=7), but the ex- studies. Annals of the New York Academy of Science 836, clusion of female subjects from data analysis did not 269–287. reduce the variability or enhance the difference be- Arora RC, Meltzer HY (1989). Serotonergic measures in the brains of suicide victims: 5-HT2 binding sites in the frontal tween patients and controls. The increased 5-HT2 re- A cortex of suicide victims and control subjects. American ceptor density in the caudate nuclei suggests an Journal of Psychiatry 146, 730–736. altered balance in the serotonin system in OCD. This Attar-Levy D, Martinot JL, Blin J, Dao-Castellana MH, is in accordance with both the serotonin hypothesis – Crouzel C, Mazoyer B, Poirier MF, Bourdel MC, Aymard stating that there is an imbalance in the serotonin N, Syrota A, Feline A (1999). The cortical serotonin2 neurotransmitter system in OCD – and also the feed- receptors studied with positron-emission tomography and back loop between thalamus and the orbito-frontal [18F]-setoperone during depressive illness and cortex, the caudate nucleus, and globus pallidus, antidepressant treatment with clomipramine. Biological regions that all are known to be involved in the path- Psychiatry 45, 180–186. ology of OCD. Thus, this loop is assumed to be crucial Baxter Jr. LR, Phelps ME, Mazziotta JC, Guze BH, Schwartz JM, Selin CE (1987). Local cerebral glucose metabolic rates for the behavioural control lost in OCD patients, and in obsessive-compulsive disorder. A comparison with down-regulation of the caudate ‘hyperactivity’ is rates in unipolar depression and in normal controls. thought to be implicated in the response to treatment Archives of General Psychiatry 44, 211–218. with SSRIs, as serotonin is assumed to have an Baxter Jr. LR, Schwartz JM, Bergman KS, Szuba MP, Guze inhibitory effect in this feedback loop (Insel, 1992). BH, Mazziotta JC, Alazraki A, Selin CE, Ferng HK, In conclusion, we find that in untreated OCD Munford P, Phelps ME (1992). Caudate glucose metabolic patients an increase in 5-HT2A receptor binding is rate changes with both drug and behavior therapy for found in the caudate nuclei. The increase in 5-HT2A obsessive-compulsive disorder. Archives of General receptors might occur as a compensation for a relative Psychiatry 49, 681–689. lack of serotonin in the feedback loop between the Baxter Jr. LR, Schwartz JM, Mazziotta JC, Phelps ME, Pahl thalamus and the orbito-frontal cortex, the caudate JJ, Guze BH, Fairbanks L (1988). Cerebral glucose metabolic rates in nondepressed patients with obsessive- nucleus and globus pallidus. Further studies are compulsive disorder. American Journal of Psychiatry 145, needed to determine whether the increase in 5-HT2 A 1560–1563. receptor binding in the caudate nuclei is representa- Biver F, Lotstra F, Monclus M, Wikler D, Damhaut P, tive for other receptors of the 5-HT family, and to what Mendlewicz J, Goldman S (1996). Sex difference in 5-HT2 extent our finding can be regarded as a trait or state receptor in the living human brain. Neuroscience Letters 204, marker. 25–28. 400 K. H. Adams et al.

Cottraux J, Note I, Yao SN, Lafont S, Note B, Mollard E, injection of radioactive tracers. Journal of Cerebral Blood Flow Bouvard M, Sauteraud A, Bourgeois M, Dartigues JF Metabolism 12, 709–716. (2001). A randomized controlled trial of cognitive therapy Lemaire C, Cantineau R, Guillaume M, Plenevaux A, versus intensive behavior therapy in obsessive-compulsive Christiaens L (1991). Fluorine-18-altanserin: a radioligand disorder. Psychotherapy and Psychosomatics 70, 288–297. for the study of serotonin receptors with PET: DeGrado TR, Turkington TG, Williams JJ, Stearns CW, radiolabeling and in vivo biologic behavior in rats. Journal Hoffman JM, Coleman RE (1994). Performance of Nuclear Medicine 32, 2266–2272. characteristics of a whole-body PET scanner. Journal of Lewellen TK, Kohlmeyer SG, Miyaoka RS, Kaplan MS, Nuclear Medicine 35, 1398–1406. Stearns CW, Schubert SF (1996). Investigation of the Delgado PL, Moreno FA (1998). Hallucinogens, serotonin performance of the general electric ADVANCE positron and obsessive-compulsive disorder. Journal of Psychoactive emission tomograph in 3D mode. IEEE Transactions in Drugs 30, 359–366. Nuclear Science 43, 2199–2206. Eastwood SL, Burnet PW, Gittins R, Baker K, Harrison PJ Lindsay M, Crino R, Andrews G (1997). Controlled trial of (2001). Expression of serotonin 5-HT(2A) receptors in the exposure and response prevention in obsessive- Downloaded from https://academic.oup.com/ijnp/article/8/3/391/910011 by guest on 24 September 2021 human cerebellum and alterations in schizophrenia. compulsive disorder. British Journal of Psychiatry 171, Synapse 42, 104–114. 135–139. Goodman WK, Price LH, Rasmussen SA, Mazure C, McGuire PK, Bench CJ, Frith CD, Marks IM, Frackowiak Fleischmann RL, Hill CL, Heninger GR, Charney DS RS, Dolan RJ (1994). Functional anatomy of obsessive- (1989). The Yale-Brown Obsessive-compulsive Scale. I. compulsive phenomena. British Journal of Psychiatry 164, Development, use, and reliability. Archives of General 459–468. Psychiatry 46, 1006–1011. Meltzer CC, Smith G, Price JC, Reynolds CF, Mathis CA, Hanes KR (1996). Serotonin, psilocybin, and body Greer P, Lopresti B, Mintun MA, Pollock BG, Ben-Eliezer dysmorphic disorder: a case report. Journal of Clinical D, Cantwell MN, Kaye W, Dekosky ST (1998). Reduced Psychopharmacology 16, 188–189. binding of [18F]altanserin to serotonin type 2A receptors in Hansen ES, Hasselbalch S, Law I, Bolwig TG (2002). The aging: persistence of effect after partial volume correction. caudate nucleus in obsessive-compulsive disorder. Brain Research 813, 167–171. Reduced metabolism following treatment with paroxetine: Meyer JH, Kapur S, Eisfeld B, Brown GM, Houle S, DaSilva a PET study. International Journal of J, Wilson A, Rafi-Tari S, Mayberg H, Kennedy S (2001). Neuropsychopharmacology 5, 1–10. The effect of paroxetine on 5-HT(2A) receptors in Hollander E, DeCaria CM, Nitescu A, Gully R, Suckow RF, depression: an [(18)F]setoperone PET imaging study. Cooper TB, Gorman JM, Klein DF, Liebowitz MR (1992). American Journal of Psychiatry 158, 78–85. Serotonergic function in obsessive-compulsive disorder. Moreno FA, Delgado PL (1997). Hallucinogen-induced relief Behavioral and neuroendocrine responses to oral m- of obsessions and compulsions. American Journal of chlorophenylpiperazine and fenfluramine in patients and Psychiatry 154, 1037–1038. healthy volunteers. Archives of General Psychiatry 49, 21–28. Moses EL, Drevets WC, Smith G, Mathis CA, Kalro BN, Hrdina PD, Demeter E, Vu TB, Sotonyi P, Palkovits M Butters MA, Leondires MP, Greer PJ, Lopresti B, Loucks (1993). 5-HT uptake sites and 5-HT2 receptors in brain of TL, Berga SL (2000). Effects of estradiol and progesterone antidepressant-free suicide victims/depressives: increase administration on human serotonin 2A receptor binding: a in 5-HT2 sites in cortex and amygdala. Brain Research 614, PET study. Biological Psychiatry 48, 854–860. 37–44. Pauls DL, Leckman JF (1986). The inheritance of Gilles de la Insel TR (1992). Toward a neuroanatomy of obsessive- Tourette’s syndrome and associated behaviors. Evidence compulsive disorder. Archives of General Psychiatry 49, for autosomal dominant transmission. New England Journal 739–744. of Medicine 315, 993–997. Jenike MA, Baer L, Summergrad P, Minichiello WE, Pauls DL, Towbin KE, Leckman JF, Zahner GE, Cohen DJ Holland A, Seymour R (1990b). Sertraline in obsessive- (1986). Gilles de la Tourette’s syndrome and obsessive- compulsive disorder: a double-blind comparison with compulsive disorder. Evidence supporting a genetic placebo. American Journal of Psychiatry 147, 923–928. relationship. Archives of General Psychiatry 43, 1180–1182. Jenike MA, Hyman S, Baer L, Holland A, Minichiello WE, Perani D, Colombo C, Bressi S, Bonfanti A, Grassi F, Buttolph L, Summergrad P, Seymour R, Ricciardi J Scarone S, Bellodi L, Smeraldi E, Fazio F (1995). [18F]FDG (1990a). A controlled trial of fluvoxamine in obsessive- PET study in obsessive-compulsive disorder. A clinical/ compulsive disorder: implications for a serotonergic metabolic correlation study after treatment. British Journal theory. American Journal of Psychiatry 147, 1209–1215. of Psychiatry 166, 244–250. Karno M, Golding JM, Sorenson SB, Burnam MA (1988). Pigott TA, Hill JL, Grady TA, L’Heureux F, Bernstein S, The epidemiology of obsessive-compulsive disorder in five Rubenstein CS, Murphy DL (1993). A comparison of the US communities. Archives of General Psychiatry 45, behavioral effects of oral versus intravenous mCPP 1094–1099. administration in OCD patients and the effect of Lassen NA (1992). Neuroreceptor quantitation in vivo by the metergoline prior to i.v. mCPP. Biological Psychiatry 33, steady-state principle using constant infusion or bolus 3–14. 5-HT2A receptor binding in patients with OCD 401

Pinborg LH, Adams KH, Svarer C, Holm S, Hasselbalch SG, Schwartz JM, Stoessel PW, Baxter Jr. LR, Martin KM, Haugbol S, Madsen J, Knudsen GM (2003). Phelps ME (1996). Systematic changes in cerebral glucose

Quantification of 5-HT2A receptors in the human brain metabolic rate after successful behavior modification using [18F]altanserin-PET and the bolus/infusion treatment of obsessive-compulsive disorder. Archives of approach. Journal of Cerebral Blood Flow Metabolism 23, General Psychiatry 53, 109–113. 985–996. Sheline YI, Mintun MA, Moerlein SM, Snyder AZ (2002). Pinborg LH, Adams KH, Yndgaard S, Hasselbaclh SG, Greater loss of 5-HT(2A) receptors in midlife than in late Holm S, Kristiansen H, Paulson OB, Knudsen GM life. American Journal of Psychiatry 159, 430–435. 18 (2004). [ F]Altanserin binding to human 5HT2A receptors Smith GS, Price JC, Lopresti BJ (1998). Test-retest variability is unaltered after citalopram and pindolol challenge. of serotonin 5HT2A receptor binding measured with Journal of Cerebral Blood Flow Metabolism 24, 1037–1045. positron emission tomography and [18F]-Altanserin in the Rauch SL, Jenike MA, Alpert NM, Baer L, Breiter HC, human brain. Synapse 30, 380–392. Savage CR, Fischman AJ (1994). Regional cerebral blood Swedo SE, Pietrini P, Leonard HL, Schapiro MB, Rettew flow measured during symptom provocation in obsessive- DC, Goldberger EL, Rapoport SI, Rapoport JL, Grady CL Downloaded from https://academic.oup.com/ijnp/article/8/3/391/910011 by guest on 24 September 2021 compulsive disorder using oxygen 15-labeled carbon (1992). Cerebral glucose metabolism in childhood-onset dioxide and positron emission tomography. Archives of obsessive-compulsive disorder. Revisualization during General Psychiatry 51, 62–70. pharmacotherapy. Archives of General Psychiatry 49, Ravindran AV, Lapierre YD, Anisman H (1999). 690–694. Obsessive-compulsive spectrum disorders: effective Swedo SE, Schapiro MB, Grady CL, Cheslow DL, Leonard treatment with paroxetine. Canadian Journal of Psychiatry HL, Kumar A, Friedland R, Rapoport SI, Rapoport JL 44, 805–807. (1989). Cerebral glucose metabolism in childhood-onset Rosier A, Dupont P, Peuskens J, Bormans G, Vandenberghe obsessive-compulsive disorder. Archives of General R, Maes M, Grott T, Schiepers C, Verbuggen A, Psychiatry 46, 518–523. Mortelmans L (1996). Visualisation of loss of 5-HT2A Tan PZ, Baldwin RM, Van Dyck CH, Al-Tikriti M, Roth B, receptors with age in healthy volunteers using Khan N, Charney DS, Innis RB (1999). Characterization of [18F]altanserin and positron emission tomographic radioactive metabolites of 5-HT2A receptor PET ligand. imaging. Psychiatry Research 68, 11–22. Nuclear Medicine and Biology 26, 601–608. Saxena S, Brody AL, Ho ML, Alborzian S, Ho MK, Talairach J, Tournoux P (1988). Co-Planar Stereotaxic Atlas of Maidment KM, Huang SC, Wu HM, Au SC, Baxter Jr. BL the Human Brain. Stuttgart, Germany: Georg Thieme (2001). Cerebral metabolism in major depression and Verlag. obsessive-compulsive disorder occurring separately and Weissman MM, Bland RC, Canino GJ, Greenwald S, Hwu concurrently. Biological Psychiatry 50, 159–170. HG, Lee CK, Newman SC, Oakley-Browne MA, Rubio- Saxena S, Brody AL, Ho ML, Alborzian S, Maidment KM, Stipec M, Wickramaratne PJ, Wittchen HC, Yeh EK Zohrabi N, Ho KM, Huang SC, Wu HM, Baxter Jr. BL (1994). The cross national epidemiology of obsessive- (2002). Differential cerebral metabolic changes with compulsive disorder. The Cross National Collaborative paroxetine treatment of obsessive-compulsive disorder vs Group. Journal of Clinical Psychiatry 55 (Suppl.), 5–10. major depression. Archives of General Psychiatry 59, 250–261. WHO (1992–1994). ICD-10: International Statistical Saxena S, Brody AL, Maidment KM, Dunkin JJ, Classification of Diseases and Related Health Problems: 10th Colgan M, Alborzian S, Phelps ME, Baxter Jr. BL (1999). Revision. Geneva: World Health Organization. Localized orbito-frontal and subcortical metabolic changes Zohar J, Judge R (1996). Paroxetine versus clomipramine in and predictors of response to paroxetine treatment in the treatment of obsessive-compulsive disorder. OCD obsessive-compulsive disorder. Neuropsychopharmacology Paroxetine Study Investigators. British Journal of Psychiatry 21, 683–93. 169, 468–74.