Symptom dimensions in obsessive-compulsive disorder

[摘要] Background: Obsessive-compulsive disorder (OCD) is a neuropsychiatric conditioncharacterized by significant heterogeneity. It has been suggested that classification of OCDinto more homogeneous subtypes, and identification of their associations with etiologicalfactors (e.g. genetic variants, or psychological trauma), and outcome (e.g. disability andtreatment response), may be useful. The identified subtypes are not definitive yet andcontinue to be subject to revision. The overall objective of this dissertation was to assesscomprehensively a sample of OCD patients, and to use cluster analytic methods to delineatevalid OCD subtypes.Methods: Patients meeting DSM-IV criteria for a primary diagnosis of OCD (N=261) on theStructured Clinical Interview for Axis I Disorders - Patient Version (SCID-I/P), with agesranging from 16 to 71, took part in the study. The newly developed Structured ClinicalInterview for the Diagnosis of putative Obsessive-Compulsive Spectrum Disorders (SCIDOCSD)was administered to assess OCD-related conditions not covered by the SCID-I/P.OCD subtyping, based on OCD symptomatology (assessed with the Yale-Brown Obsessive-Compulsive Symptom Checklist [YBOCS-CL]), and based on comorbidity with the OCDspectrum of disorders (assessed with the SCID-OCSD), proceeded along the following lines:Firstly, latent class cluster analysis (LCA), a categorical analogue to traditional factoranalysis (FA), and with many advantages compared to FA, was implemented with the (nine)most frequently endorsed OC symptoms. Secondly, in an attempt to remedy some of thelimitations of the LCA (e.g. increased potential for computational instability when additionalindicators / symptoms were included), cluster analyses (Ward's method) were performed onall of the items of the YBOCS-CL and SCID-OCSD, respectively, for all OCD patients. Theassociations of cluster scores with demographic variables (age, gender), clinical variables(age of onset, obsessive-compulsive symptom severity and dimensions, level of insight,temperament, childhood trauma, treatment response) and genotypes were then examined, using Spearman correlation coefficients, one-way analysis of variance (ANOVA), and Mann-Whitney U-tests, where appropriate.Results: The findings suggested that increased presentation of symptoms characteristic ofeach of the clusters of cases was associated with specific demographic and clinicalcharacteristics, which substantiated the presence of distinct clinical subtypes of OCD.Cluster analysis of the 45 selected items of the YBOCS-CL in this sample of OCD patientsidentified 6 separate clusters; these clusters were labelled 'Contamination fears / washing,'Hoarding / collecting, 'Symmetry / ordering / counting / arranging / repeating, 'Sexual,'Somatic, religious and diverse and 'Harm-related. Increased presentation of symptomscharacteristic of each of the clusters was associated with specific demographic, clinical and,in some cases, genetic characteristics. Of note, the findings indicated the L/L (met/met)genotype of COMT Val158Met polymorphism plays a major role in the increasedmanifestation of sexual, somatic, religious and diverse, and harm-related symptoms of OCD,as such contributing to the relatively limited data on OC symptom subtypes and genetics.However, the fact that the associated features did not clearly and uniquely differentiateclusters and that clusters were significantly correlated with one another suggested that thedelineation of the OCD complex into OC symptom clusters is not the only way to approachthe heterogeneity characteristic of OCD. Nevertheless, the significant comorbidity withOCSD's in the identified clusters (e.g. tics associated with sexual obsessions / compulsions)highlighted the significant relationship of OCD with the OCSD's. This again raised thequestion about the way in which the OCSD's 'fit with the standard OC symptomatologyoutlined in the YBOCS-CL. A cluster analysis of OCSD's in OCD patients identified aTourette's syndrome / tics subtype of OCD (part of the so-called 'reward deficiency cluster),as well as an impulsivity subtype, and a somatic subtype – each associated with specificclinical and demographic variables. Here, a significant relationship between the identified clusters and the investigated dopaminergic and serotonergic polymorphisms was not found,suggesting that variants in other genes in these systems should also be explored.Conclusion: The main finding was that OCD is indeed a heterogeneous disorder that maybe subtyped into different symptom dimensions. The identified OCD subtypes with theirassociated features were to a large extent consistent with previously published data.However, in contrast to factor analysis, LCA provided a novel, appropriate andadvantageous statistical analysis strategy for the data. Furthermore, to our knowledge, theattempt to classifiy OCD according to comorbid OCSD's was the first cluster analysis basedon a prospective comprehensive interview investigating a range of OCSD's. As such,although the dimensional structure of OCD is still not entirely understood, the categorizationof our OCD patients into different groups and the investigation of their respective featureshave gone beyond the literature and thus add another dimension to the increasing efforts tofully delineate OCD subtypes.

[发布机构] Stellenbosch University