* Understand how a factor-analytic approach can be used in conjunction with neuroimaging to study psychiatric diseases
* Discuss current theories about the brain regions that may be involved in OCD
* Learn which OCD symptoms tend to appear in clusters

Abstract

Various schemes have been developed in attempts to define meaningful subtypes of obsessive-compulsive disorder (OCD). A factor-analytic approach presumes that the symptoms of the disorder can be described by several independent symptom dimensions. The purpose of this study was to explore the neural correlates of three symptom dimensions that were derived from previous factor analyses. Positron emission tomography was used to measure relative regional cerebral blood flow (rCBF) in 14 subjects with OCD while they engaged in a continuous performance task. Clinical indices, including factor scores, were ascertained via structured interviews plus administration of the Yale-Brown Obsessive-Compulsive Scale and checklist. The severity of Factor 1 (religious/aggressive/sexual obsessions and checking) was positively correlated with rCBF in striatum bilaterally. In addition, distinct trends were observed for the other two Factors. These findings provide initial support for a modular neurobiologic model of OCD, where dysfunction within separate component systems may principally mediate independent symptom factors. More important, this novel strategy may represent a powerful new approach to interpreting brain imaging studies of neuropsychiatric diseases.
Introduction

Neuroimaging studies performed to date using various paradigms have yielded convergent findings in the pathophysiology of obsessive-compulsive disorder (OCD), implicating involvement of the orbitofrontal and anterior cingulate cortices, as well as the striatum and thalamus.1-3 The failure to replicate results perfectly from one experiment to another may be attributable to differences in the constitution of the groups studied, or to other methodologic differences. Since psychiatric diagnoses in general, and OCD in particular, represent syndromes (ie, constellations of signs and symptoms), it is appealing to consider that the conditions as defined may be further subdivided, or characterized in a more refined manner, to reflect discrete underlying pathophysiologic processes. The idea of a pathophysiology-based nomenclature represents an attractive model4 that is consonant with the principal goal of developing diagnostic designations that predict the natural course of a condition or response to treatment, and thereby serve to better inform clinical management of individual patients.

In the case of OCD, there is no question that substantial clinical heterogeneity exists, both in terms of the variety of obsessions and compulsions, and in the range of clinical responses to various therapies. Initial attempts to subdivide OCD phenomenologically focused on the distinctions between obsessions and compulsions (ie, "checkers" vs "washers").5,6 Subsequently, the distinction between OCD with comorbid tics vs OCD without such tics became popular.7-11

More recently, several investigators have taken a different approach that seeks empirically to identify independent factors that, taken together, comprise the full picture of OCD.12,13 This method of "factor analysis" has been used to identify independent groups (or clusters) of highly intercorrelated OCD symptoms, and is a strategy that has yielded robust results over several studies that include data from over 500 patients12,13 (D. Mataix-Cols, PhD, et al, unpublished data, 1998). Collectively, these analyses suggest that OCD can be conceptualized as the product of at least three independent factors (Table 1): Factor 1: checking compulsions and religious, aggressive, and sexual obsessions; Factor 2: symmetry and ordering symptoms; and Factor 3: washing and cleaning symptoms12,13 (D. Mataix-Cols, PhD, et al, unpublished data, 1998). Factors 1 and 2 notably were associated with comorbid tics in one study.13 In other studies, Factor 2 alone was associated with comorbid tics12 and superior response to cingulotomy.14

It is important to appreciate the subtle ways in which this factor-analytic approach differs from more conventional categorical subtyping of disease. By way of analogy, first consider the case of pneumonia, a classic example of a syndromic diagnosis (ie, characterized by fever, cough, lung infiltrates visualized by chest x-ray, etc). Upon discovery of the etiological and pathophysiological underpinnings of pneumonia, this syndrome can be subdivided into gram-positive, gram-negative, or nonbacterial types based on characteristics of the offending agents. These designations are categorical, and essentially mutually exclusive. Moreover, each suggests a different prescription for definitive medical intervention. By contrast, consider the case of a sprained knee, also initially a syndromic diagnosis (ie, characterized by pain, tenderness, and functional disability following mechanical trauma to the joint). Upon further examination, one could detect independent involvement of any combination of the ligaments normally serving to stabilize the joint. Interestingly, one might assume (at least for the purposes of this illustration) that damage to the medial collateral (MCL), lateral collateral (LCL), anterior cruciate, or posterior cruciate ligaments can occur to varying severities independent of one another. Hence, some patients might have minor involvement of the MCL and severe involvement of the LCL and preserved integrity of the others, or minor damage to all four, and so on. Of course, the interplay of the pathology among the different components of the system will necessarily translate into different clinical pictures, and a need for different interventions. (For instance, surgical repair of the MCL would obviously not be indicated if only the LCL were damaged.) This orthopedic analogy may ring hollow given our sense of the complexity and interdependence of brain systems. However, there are precedents for similar modular theories of brain diseases as well. A simplified interpretation of the Geschwind model of language and aphasias,15 for instance, suggests that dysfunction within anterior components of the system (ie, Broca¹s area) can produce expressive aphasias independent of receptive aphasic symptoms produced by lesions of the posterior components of the system (eg, Wernicke¹s area). Moreover, a third type of aphasia can occur owing to disconnection of the anterior and posterior components.

Which model, then, is most appropriate for OCD? The factor-analytic approach serves as a basis for characterizing different dimensions of OCD independently, in contrast to the Yale Brown Obsessive-Compulsive Scale (Y-BOCS),16,17 which seeks to provide an integrated assessment of global symptom severity. Given the reliability of recent findings using the factor-analytic approach, our group has initiated studies that seek to determine correlates of these OCD-symptom factors, both in terms of treatment response (D. Mataix-Cols, PhD, et al, unpublished data, 1998) and brain-activity profiles. In the current experiment, we report preliminary findings on positron emission tomography (PET) scanning regarding the neural correlates of these factor-analyzed symptom dimensions.

A priori, we posed two sets of hypotheses: (A) Factors 1 and 2 might be related to striatal and/or thalamic involvement, given the purported role of striatothalamic circuitry in the OCD-tic spectrum3,18,19 and the association of Factors 1 and 2 with tics.12,13 Moreover, hippocampal involvement might be associated with striatal dysfunction, and thus these same factors, based on findings of a recent cognitive activation study of OCD.19 (B) Factor 3 might be related to orbitofrontal and anterior cingulate (ie, prefrontal) cortical involvement, given the purported target of serotonin reuptake inhibitors (SRIs) within the frontal cortex20 and the suggestion that OCD without tics is more SRI-responsive than OCD with tics.7,10
Subjects and Methods

All procedures were conducted in accordance with the Subcommittee on Human Studies of the Massachusetts General Hospital. Written informed consent was obtained from all subjects prior to participation.

Fourteen patients with OCD (Table 2) were recruited via the Massachusetts General Hospital OCD Clinic and Research Unit as paid subjects. All subjects were right-handed.21 A diagnosis of OCD was made by psychiatric examination and confirmed by structured clinical interview.22 Participants had no history of psychosis, substance dependence, bipolar disorder, or current substance abuse. They were medically healthy by self report and had no history of significant head injury, seizure, or neurologic or current major medical conditions (in particular, comorbid tics were a basis for exclusion). Finally, according to the patient histories, no subject had taken any psychotropic medications during the 4 weeks prior to testing, nor any other medicines that would interfere with the study procedures.

Clinical assessment included administration of the Y-BOCS as well as the Y-BOCS symptom checklist.16,17 In this manner, overall OCD severity (total Y-BOCS score) and an inventory of specific current OCD symptoms were ascertained. Clinical interviews were also conducted, which enabled ascertainment of OCD symptoms that were currently prominent vs mild for each subject. Using the factor structure defined by Leckman et al,13 each subject was assigned a score of 0 (absent), 1 (mild), or 2 (prominent) for each factor. The value assigned for each factor reflected the highest score for any of the elements comprising that factor (Table 1).

The PET data to be analyzed were acquired as part of a separate study, the initial findings of which have been reported elsewhere.19 Specifically, PET data were acquired while subjects performed a simple continuous performance task, as the first functional run after being positioned in the scanner. The continuous performance task was the baseline condition from the serial reaction time task.19,23,24 The paradigm entails serial presentations of visual cues (ie, asterisks), one at a time, at any of four spatial locations displayed on a computer monitor, with an interstimulus interval of 1 second. Four keys were positioned beneath the monitor, with one key corresponding to each of the potential stimulus locations. Subjects covered these four keys with the first two fingers on each hand, and were instructed to respond as quickly and accurately as possible to each stimulus presentation by pressing the corresponding key with the corresponding finger. Stimuli were presented in blocks of 144 trials. For the current study, the order of the stimulus locations was random. This represents a classic continuous performance task with visuospatial, motor, and attentional demands. Each subject performed three blocks of trials; the first two blocks represented practice sessions, and the third was performed during PET data acquisition. The PET data were acquired via a Scanditronix PC4096 PET camera (General Electric, Milwaukee, Wis) while subjects inhaled 15O-labeled-CO2 for 1 minute. Following reconstruction, movement-corrected,25 whole-brain normalized images reflecting relative rCBF were transformed to Talairach space.26,27 After spatial normalization, scans were filtered with a 20-mm (full width at half maximum), two-dimensional gaussian filter.

The PET data analysis followed the theory of statistical parametric mapping,28-31 and was performed with SPM95 software (Wellcome Dept of Cognitive Neurology, London, UK). Across the entire cohort of subjects, a total of four statistical parametric maps were generated: one map for each symptom factor (Factors 1, 2, and 3) and one map for the overall Y-BOCS score. For each map, the covariates-only option was selected, so that a regression analysis was performed to test the linear relationship between rCBF and the external clinical variable (across subjects), yielding a z score at each voxel in space. For ease of discussion, we refer to the findings in terms of significant "correlations," although the analysis employed formally involved linear regression, rather than assessment of correlation per se.

The statistical parametric maps were inspected to identify foci that had significant correlations within the five predetermined regions of interest (ROIs): orbitofrontal cortex, anterior cingulate cortex, striatum, thalamus, and hippocampus. The boundaries of the relevant search volumes were defined stereotactically, as were the nominal locations of all identified significant correlation foci. Since this represents the first study of this type, we chose to employ relatively liberal statistical thresholds; more stringent thresholds and more circumscribed hypotheses are recommended for follow-up experiments. Although a [zeta] score of >3.09 (uncorrected for multiple comparisons) was selected as our threshold for statistical significance, we report all loci within predetermined ROIs corresponding to [zeta]>1.96 (P<0.05, uncorrected for multiple comparisons). Given the fact that the core analysis entailed 15 separate comparisons (five ROIs with three factors at each ROI), and that each ROI encompassed multiple resolution units, this statistical threshold is admittedly liberal. To obviate bias, we also report loci outside of ROIs where [zeta]>3.09; however, in the absence of hypotheses pertaining to those territories, such findings should not be taken as strong evidence of reliable effects.
Results

Demographic and clinical data, as well as factor scores, are summarized in Table 2. Note that this cohort yielded a relatively balanced distribution of scores across all three Factors, as well as a broad range of total Y-BOCS scores; such distributions are essential for generating adequate range for regression analyses.

The imaging results are presented in Table 3 and Figure 1. Significant positive correlations were found between Factor 1 scores and rCBF within the striatum bilaterally ([zeta]=3.70 [right] and 3.28 [left]; see Fig 1), but not in any other designated ROIs. Post hoc analysis revealed a negative correlation within the right parietal cortex ([zeta]=3.43). The analysis of Factor 2 yielded only a trend toward negative correlation within the right striatum ([zeta]=2.84), but not in any other designated ROI. Post hoc analysis demonstrated negative correlations within the left middle temporal gyrus ([zeta]=3.19) and left temporo-insular cortex ([zeta]=3.17). Factor 3 analysis showed a trend toward positive correlations within the bilateral anterior cingulate cortex (both left and right [zeta]=2.13) and left orbitofrontal cortex ([zeta]=2.25), but not in any other designated ROI. Post hoc analysis revealed positive correlations within the right dorsolateral prefrontal cortex (~Brodmann area 46; [zeta]=3.60), left postcentral gyrus ([zeta]=3.55), left midcingulate cortex (~Brodmann area 31; [zeta]=3.35), and left superior temporal gyrus ([zeta]=3.20), as well as a negative correlation within the left visual cortex ([zeta]=3.98).

The analysis of total Y-BOCS scores indicated that none of the above findings was confounded by correlation with overall severity of illness. In fact, no significant correlations were observed between rCBF and total Y-BOCS scores within the designated ROIs. Post hoc analysis revealed a positive correlation within the right visual cortex ([zeta]=3.90) and a negative correlation within the right insular cortex ([zeta]=3.16).
Discussion

To our knowledge, this represents the first PET study of its kind that attempts to identify neural correlates of clinical dimensions in OCD as defined by factor analysis. The findings here should be viewed as preliminary. In fact, the primary goal of this project was to illustrate how this novel strategy may be applied in future psychiatric neuroimaging studies.

Nonetheless, the current results are intriguing in their own right. A significant positive correlation was observed within bilateral striatum for Factor 1 (religious/aggressive/sexual obsessions and checking). This indicated that higher levels of rCBF were associated with greater severity of OCD symptoms within this Factor. The anatomic extent of this finding was essentially pan-striatal, and included areas of caudate nucleus, putamen, and pallidum. In contrast, Factor 2 (symmetry and ordering) was associated with a trend toward negative correlation within the right caudate nucleus. Generally, the implication that activity within striatal regions is associated with symptom severity for the two Factors previously associated with tics is consistent with our hypotheses. The finding that Factor 3 was associated with a trend toward positive correlations within several prefrontal areas, including the left orbitofrontal cortex, bilateral anterior cingulate cortex, and right dorsolateral prefrontal cortex, is also consistent with our predictions. It is noteworthy that none of these findings was confounded by the overall Y-BOCS scores. The fact that the Y-BOCS analysis failed to yield significant correlations within any of the predetermined ROIs further supports the contention that clinical heterogeneity in functional neuroimaging studies may obscure the detection of relevant neural correlates of OCD.

At first glance, the finding that one factor (Factor 1) is positively correlated and another (Factor 2) negatively correlated with striatal activity might seem contradictory. To the contrary, it is quite plausible that these Factors may reflect differing forms of striatal dysfunction. In fact, it has been hypothesized that the symptoms of OCD could be mediated either by hyperactivity within the direct striato-pallido-thalamic pathway (here consistent with the positive correlation with Factor 1) or hypoactivity within the indirect pathway (consistent with the negative correlation with Factor 2).18,32 Indeed, these reciprocal forms of pathophysiology may explain the inconsistency in findings of striatal abnormalities in neutral state studies of OCD, wherein symptom subtypes have typically not been controlled and only rarely even ascertained or reported. Similarly, it may be that the literature has been most consistent with respect to orbitofrontal and anterior cingulate activations precisely because Factor 3-rich study populations have predominated.

It is premature to seek explanations for every trend reported above. However, with these concepts in mind, it may be illuminating for investigators to revisit previously gathered data sets. For instance, in our own PET symptom provocation study of OCD,33 we recruited six patients with primarily Factor 3 symptoms and two patients with primarily Factor 1 symptoms. In that experiment, we were able to demonstrate recruitment of orbitofrontal cortex, anterior cingulate cortex, and striatum; however, only left orbitofrontal cortical activation correlated positively with the magnitude of OCD symptom provocation. Of note is the fact that comparable dorsolateral prefrontal activation was also observed. In retrospect, this constellation of results appears consonant with our current findings, especially in light of the population studied.

In this context, it is important to acknowledge the potential influence of state variables. In the current study we analyzed data acquired during a nominally neutral state‹specifically, during a simple continuous performance task. It would be interesting to conduct analogous factor-analysis-guided interpretations of symptom provocation, pre- and posttreatment, and cognitive activation data, as well as receptor characterization studies.

Several limitations of the current study should be noted. As is typical in functional imaging research, the number of subjects was modest; consequently, the results are vulnerable to statistical errors of types 1 and 2. Other limitations include those intrinsic to the imaging methods employed.34 For instance, localization of statistically significant foci is constrained by the spatial resolution of PET, and further complicated by normalization to Talairach space. Here, the reliance on correlation analysis places a special burden on the ascertainment of symptom severity. We employed a crude scoring system on a 0­2 point scale. Future studies of this type might utilize separate full-scale 40-point Y-BOCS assessments for the elements of each Factor. This could provide a broader range of scores with greater sensitivity and validity across cohorts.

The strengths of this work include a well-characterized study population, well-standardized acquisition parameters, and an innovative approach to the treatment of PET data analysis. We anticipate that future studies of OCD, as well as other neuropsychiatric disorders, might employ this or similar strategies for identifying neural correlates of independent symptom factors. We also recommend applying this approach to the analysis of treatment outcomes; as mentioned above, at least one such project is already under way (D. Mataix-Cols, PhD, et al, unpublished data, 1998).

Although hope exists that functional neuroimaging tools may someday provide useful data for the routine clinical management of OCD, it would be more cost-effective if easily accessible indices (such as overt symptom information) could serve this purpose instead. In the meantime, these findings provide initial support for a modular neurobiologic model of OCD, whereby dysfunction occurring within separate component systems may principally mediate independent symptom factors. Additional experiments will be necessary to replicate and expand upon these preliminary results.
References

1. Insel TR. Toward a neuroanatomy of obsessive-compulsive disorder. Arch Gen Psychiatry. 1992;49:739-744.
2. Hoehn-Saric R, Benkelfat C. Structural and functional brain imaging in obsessive compulsive disorder. In: Hollander E, Zohar J, Marazziti D, et al, eds. Current Insights in Obsessive Compulsive Disorder. New York, NY: John Wiley and Sons; 1995:183-214.
3. Rauch SL, Bates JF, Grachev ID. Obsessive compulsive disorder. Child Adolesc Psychiatr Clin Am. 1997;6:365-381.
4. Rauch SL. Advances in neuroimaging research: how might they influence our diagnostic classification scheme? Harv Rev Psychiatry. 1996;4:159-162.
5. Rachman S, Hodgson R. Obsessions and Compulsions. Englewood Cliffs, NJ: Prentice Hall; 1980.
6. Baer L, Jenike MA. Introduction. In: Jenike MA, Baer L, Minichiello WE, eds. Obsessive Compulsive Disorders: Theory and Management. 2nd ed. Chicago, IL: Year Book Medical Publishing; 1990:3-9.
7. McDougle CJ, Goodman WK, Price LH, et al. Neuroleptic addition in fluvoxamine refractory obsessive compulsive disorder. Am J Psychiatry. 1990;147:652-654.
8. George MS, Trimble MR, Ring HA, et al. Obsessions in obsessive-compulsive disorder with and without Gilles de la Tourette's syndrome. Am J Psychiatry. 1992;150:93-97.
9. Holzer JC, Goodman WK, McDougle CJ, et al. Obsessive compulsive disorder with and without a chronic tic disorder: a comparison of symptoms in 70 patients. Br J Psychiatry. 1994;164:469-473.
10. McDougle CJ, Goodman WK, Leckman JF, et al. Haloperidol addition in fluvoxamine-refractory obsessive-compulsive disorder: a double-blind, placebo-controlled study in patients with and without tics. Arch Gen Psychiatry. 1994;51:302-308.
11. Miguel EC, Baer L, Coffey BJ, et al. Phenomenological differences appearing with repetitive behaviors in obsessive-compulsive disorder and Gilles de la Tourette's syndrome. Br J Psychiatry. 1997;170:140-145.
12. Baer L. Factor analysis of symptom subtypes of obsessive compulsive disorder and their relation to personality and tic disorders. J Clin Psychiatry. 1994;55:18-23.
13. Leckman JF, Grice DE, Boardman J, et al. Symptoms of obsessive-compulsive disorder. Am J Psychiatry. 1997;154:911-917.
14. Baer L, Rauch SL, Ballantine HT, et al. Cingulotomy for intractable obsessive-compulsive disorder: prospective long-term follow-up of 18 patients. Arch Gen Psychiatry. 1995;52:384-392.
15. Geschwind N. The organization of language and the brain. Science. 1970;170:940-944.
16. Goodman WK, Price LH, Rasmussen SA, et al. The Yale-Brown Obsessive Compulsive Scale (Y-BOCS), part I: development, use, and reliability. Arch Gen Psychiatry. 1989;46:1006-1011.
17. Goodman WK, Price LH, Rasmussen SA, et al. The Yale-Brown Obsessive Compulsive Scale (Y-BOCS), part II: validity. Arch Gen Psychiatry. 1989;46:1012-1016.
18. Rauch SL, Jenike MA. Neural mechanisms of obsessive-compulsive disorder. Current Review of Mood and Anxiety Disorders. 1997;1:84-94.
19. Rauch SL, Savage CR, Alpert NM, et al. Probing striatal function in obsessive compulsive disorder: a PET study of implicit sequence learning. J Neuropsychiatry. 1997;9:568-573.
20. Mansari ME, Bouchard C, Blier P. Alteration of serotonin release in the guinea pig orbito-frontal cortex by selective serotonin reuptake inhibitors: relevance to treatment of obsessive-compulsive disorder. Neuropsychopharmacology. 1995;13:117-127.
21. Oldfield RC. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia. 1971;9:97-113.
22. Spitzer RL, Williams JBW, Gibbon M, et al. Structured Clinical Interview for DSM-III-R (SCID). Washington, DC: American Psychiatric Press; 1990.
23. Nissen MJ, Bullemer P. Attentional requirements of learning: evidence from performance measures. Cognitive Psychology. 1987;19:1-32.
24. Rauch SL, Savage CR, Brown HD, et al. A PET investigation of implicit and explicit sequence learning. Hum Brain Mapping. 1995;3:271-286.
25. Alpert NM, Berdichevsky D, Levin Z, et al. Improved methods for image registration. Neuroimage. 1996;3:10-18.
26. Talairach J, Tournoux P. Co-Planar Stereotaxic Atlas of the Human Brain. New York, NY: Thieme Medical Publishers Inc; 1988.
27. Alpert NM, Berdichevsky D, Weise S, et al. Stereotactic transformation of PET scans by nonlinear least squares. In: Uemura K, Lassen NA, Jones T, eds. Quantification of Brain Function: Tracer Kinetics and Image Analysis in Brain PET. Amsterdam, The Netherlands: Elsevier; 1993:459-463.
28. Friston KJ, Frith CD, Liddle PF, Frackowiak RSJ. Comparing functional (PET) images: the assessment of significant change. J Cereb Blood Flow Metab. 1991;11:690-699.
29. Worsley KJ, Evans AC, Marrett S, et al. A three dimensional statistical analysis for rCBF activation studies in human brain. J Cereb Blood Flow Metab. 1992;12:900-918.
30. Friston KJ, Worsley KJ, Frackowiak RSJ, et al. Assessing the significance of focal activations using their spatial extent. Hum Brain Mapping. 1994;1:214-220.
31. Friston KJ, Holmes AP, Worsley KJ, et al. Statistical parametric maps in functional imaging: a general approach. Hum Brain Mapping. 1995;2:189-210.
32. Baxter LR, Schwartz JM, Guze BH, et al. Neuroimaging in obsessive-compulsive disorder: seeking the mediating neuroanatomy. In: Jenike MA, Baer L, Minichiello WE, eds. Obsessive Compulsive Disorder: Theory and Management. 2nd ed. Chicago, Ill: Year Book Medical Publishers; 1990:167-188.
33. Rauch SL, Jenike MA, Alpert NM, et al. Regional cerebral blood flow measured during symptom provocation in obsessive-compulsive disorder using 15O-labeled CO2 and positron emission tomography. Arch Gen Psychiatry. 1994;51:62-70.
34. Rauch SL, Savage CR, Alpert NM, et al. A positron emission tomographic study of simple phobic symptom provocation. Arch Gen Psychiatry. 1995;52:20-28.

* Dr. Rauch is director of psychiatric neuroimaging research in the Departments of Psychiatry and Radiology at Massachusetts General Hospital in Boston, and associate professor of psychiatry at Harvard Medical School in Cambridge, MA.
* Dr. Dougherty is clinical assistant in psychiatry at Massachusetts General Hospital in Boston, and clinical fellow in psychiatry at Harvard Medical School in Cambridge, MA.
* Dr. Shin is research fellow in psychology at Massachusetts General Hospital in Boston and Harvard Medical School in Cambridge, MA.
* Dr. Alpert is director of the PET Imaging Laboratory in the Department of Radiology at Massachusetts General Hospital in Boston, and associate professor of radiology at Harvard Medical School in Cambridge, MA.
* Mr. Manzo is research assistant in the Department of Psychiatry at Massachusetts General Hospital in Boston, MA.
* Ms. Leahy is research coordinator in the Department of Psychiatry at Massachusetts General Hospital in Boston, MA.
* Dr. Fischman is chief of nuclear medicine in the Department of Radiology at Massachusetts General Hospital in Boston, and associate professor of radiology at Harvard Medical School in Cambridge, MA.
* Dr. Jenike is director of the Psychiatric Neuroscience Program and associate chief of the Department of Psychiatry at Massachusetts General Hospital in Boston, and professor of psychiatry at Harvard Medical School in Cambridge, MA.
* Dr. Baer is director of research for the Obsessive-Compulsive Disorders Clinic and Research Unit in the Department of Psychiatry at Massachusetts General Hospital in Boston, and associate professor of psychology at Harvard Medical School in Cambridge, MA.